BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:In-Profile Monitoring for Multivariate Process Data in Advanced Ma
nufacturing
DTSTART:20210913T144500Z
DTEND:20210913T151500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-173@conferences.enbis.org
DESCRIPTION:Speakers: Kaibo Wang (Tsinghua University)\, chen zhang\, Du J
uan (Hong Kong University of Science and Technology)\, Peiyao Liu (Tsinghu
a University)\n\nNowadays advanced sensing technology enables real-time da
ta collection of key variables during manufacturing\, which are referred t
o as multi-channel profiles. These data facilitate in-process monitoring a
nd anomaly detection\, which have been extensively studied in the past few
years. However\, all current studies treat each profile as a whole\, such
as a high-dimensional vector or a function\, and construct monitoring sch
emes accordingly. This leads to two limitations. First\, long detection de
lay exists\, especially if the anomaly occurs in early sensing points of t
he profile. Second\, analyzing a profile as a whole requires that profiles
of different samples should be synchronized with the same length\, yet th
ey usually have certain variability due to inherent fluctuations. To addre
ss this problem\, this paper is the first to monitor multi-channel profile
s on the fly. It can not only detect anomalies without the whole profile\,
but also handle the non-synchronization effect of different samples. In p
articular\, our work is built upon the state space model (SSM) framework.
To better describe the between-state and between-profile correlations\, we
further develop the regularized SSM (RSSM). The regularizations are impos
ed as prior information\, and maximum a posterior (MAP) inference in the B
ayesian framework is adopted for parameter learning. Built upon RSSM\, a m
onitoring statistic based on one-step-ahead forecasting error is construct
ed for in-profile monitoring. The effectiveness and applicability of the p
roposed monitoring scheme are demonstrated in both the numerical studies a
nd two real case studies.\n\nhttps://conferences.enbis.org/event/11/contri
butions/173/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/173/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A novel online PCA algorithm for large variable space dimensions
DTSTART:20210915T104000Z
DTEND:20210915T110000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-198@conferences.enbis.org
DESCRIPTION:Speakers: Philipp Froehlich (University of Wuerzburg)\, Rainer
Göb\n\nPrincipal component analysis (PCA) is a basic tool for reducing t
he dimension of a space of variables. In modern industrial environments la
rge variable space dimensions up to several thousands are common\, where d
ata are recorded live in high time resolution and have to be analysed with
out time delay. Classical batch PCA procedure start from the full covarian
ce matrix and construct the exact eigenspace of the space defined by the c
ovariance matrix. The latter approach is infeasible under large dimensions
\, and even if feasible live updating of the PCA is impossible. Several so
-called online PCA algorithms are available in the literature who try to h
andle large dimensions and live updating with different approaches. The pr
esent study compares the performance of available online PCA algorithms an
d suggests a novel online PCA algorithm. The algorithm is derived by solvi
ng a simplified maximum trace problem where the optimisation is restricted
on the curve on the unit sphere\, which directly connects the respective
old principal component estimation with a projection of the newly observed
data point. The algorithm scales linearly in runtime and in memory with t
he data dimension. The advantage of the novel algorithm lies in providing
exactly orthogonal vectors whereas other algorithms lead to approximately
orthogonal vectors. Nevertheless\, the runtime of the novel algorithm is n
ot worse and sometimes even better than the one of existing online PCA alg
orithms.\n\nhttps://conferences.enbis.org/event/11/contributions/198/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/198/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Bayesian Designs for Progressively Type-I Censored Simple Step-Str
ess Accelerated Life Tests Under Cost Constraint and Order-Restriction
DTSTART:20210914T154500Z
DTEND:20210914T160500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-179@conferences.enbis.org
DESCRIPTION:Speakers: David Han\, Crystal Wiedner (The University of Texas
at San Antonio)\n\nIn this work\, we investigate order-restricted Bayesia
n cost constrained design optimization for progressively Type-I censored s
imple step-stress accelerated life tests with exponential lifetimes under
continuous inspections. Previously we showed that using a three-parameter
gamma distribution as a conditional prior ensures order restriction for pa
rameter estimation and that the conjugate-like structure provides computat
ional simplicity. Adding on to our Bayesian design work\, we explore incor
porating a cost constraint to various criteria based on Shannon informatio
n gain and the posterior variance-covariance matrix. We derive the formula
for expected termination time and expected total cost and propose estimat
ion procedures for each. We conclude with results and a comparison of the
efficiencies for the constrained vs. unconstrained tests from an applicati
on of these methods to a solar lighting device dataset.\n\nhttps://confere
nces.enbis.org/event/11/contributions/179/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/179/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Explainable AI in preprocessing
DTSTART:20210915T100000Z
DTEND:20210915T102000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-122@conferences.enbis.org
DESCRIPTION:Speakers: Paolo Giudici\, Golnoosh Babaei\, Emanuela Raffinett
i\n\nThe use of eXplainable Artificial Intelligence (XAI) in many fields\,
especially in finance has been an important issue not only for researcher
s but also for regulators and beneficiaries. In this paper\, despite recen
t researches in which XAI methods are utilized for improving the explainab
ility and interpretability of opaque machine learning models\, we consider
two mostly used model-agnostic explainable approaches namely\, Local Inte
rpretable Model Agnostic Explanations (LIME) and SHapley Additive exPlanat
ions (SHAP) as preprocessors and try to understand if the application of X
AI methods for preprocessing could improve machine learning models or not.
Moreover\, we make a comparison between the mentioned XAI methods to unde
rstand which performs better for this purpose in a decision-making framewo
rk. To validate the proposed decomposition\, we use the Lending Club\, a P
eer-to-Peer lending platform in the US\, dataset which is a reliable datas
et containing information of individual borrowers.\n\nhttps://conferences.
enbis.org/event/11/contributions/122/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/122/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Calibrating Prediction Intervals for Gaussian Processes using Cros
s-Validation method
DTSTART:20210915T102000Z
DTEND:20210915T104000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-127@conferences.enbis.org
DESCRIPTION:Speakers: Antoine Bertoncello (TotalEnergies SE)\, ACHARKI Nao
ufal\, Josselin Garnier (CMAP - Ecole Polytechnique)\n\nGaussian Processes
are considered as one of the most important Bayesian Machine Learning met
hods (Rasmussen and Williams [1]\, 2006). They typically use the Maximum L
ikelihood Estimation or Cross-Validation to fit parameters. Unfortunately\
, these methods may give advantage to the solutions that fit observations
in average (F. Bachoc [2]\, 2013)\, but they do not pay attention to the c
overage and the width of Prediction Intervals. This may be inadmissible\,
especially for systems that require risk management. Indeed\, an interval
is crucial and offers valuable information that helps for better managemen
t than just predicting a single value.\n\nIn this work\, we address the qu
estion of adjusting and calibrating Prediction Intervals for Gaussian Proc
esses Regression. First we determine the model's parameters by a standard
Cross-Validation or Maximum Likelihood Estimation method then we adjust th
e parameters to assess the optimal type II Coverage Probability to a nomin
al level. We apply a relaxation method to choose parameters that minimize
the Wasserstein distance between the Gaussian distribution of the initial
parameters (Cross-Validation or Maximum Likelihood Estimation) and the pro
posed Gaussian distribution among the set of parameters that achieved the
desired Coverage Probability.\n\nReferences :\n1. Rasmussen\, C.E.\, Wi
lliams\, C.K.I.: Gaussian Processes for Machine Learning (Adaptive C
omputation and Machine Learning). The MIT Press (2005).\n2. Bachoc\, F.:
Cross validation and maximum likelihood estimations of hyper-para
meters of gaussian processes with model misspecification. Computati
onal Statistics & Data Analysis66\, 55–69 (2013).\n\nhttps://conferences
.enbis.org/event/11/contributions/127/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/127/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Robust bootstraped h and k Mandel’s statistics for outlier detec
tion in Interlaboratory Studies
DTSTART:20210915T124000Z
DTEND:20210915T130000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-228@conferences.enbis.org
DESCRIPTION:Speakers: Cristian Solorzano (Escuela Politécnica Nacional)\,
Génesis Moreno (Escuela Politécnica Nacional)\, Miguel Alfonso Flores
Sánchez (Grupo MODES\, SIGTI\, FADE\, Departamento de Matemática\, Escu
ela Politécnica Nacional)\, Javier Tarrío Saavedra (Grupo MODES\, CITIC\
, ITMATI\, Department of Mathematics\, Escola Politécnica Superior\, Univ
ersidade da Coruña)\, Salvador Naya (MODES\, CITIC\, ITMATI\, Universidad
e da Coruña\, Escola Politécnica Superior)\n\nA new methodology based on
bootstrap resampling techniques is proposed to estimate the distribution
of the h and k Mandel's statistics\, commonly applied to identify laborato
ries that supply inconsistent results usually utilized to detect those out
lier laboratories by testing the hypothesis of reproducibility and repeata
bility (R & R)\, in the framework of Interlaboratory Studies (ILS). \n\nTr
aditionally\, the statistical tests involved in the ILS have been develope
d under theoretical assumptions of normality in the study variables. Then\
, if the variable measured by the laboratories is far from being assumed n
ormal distributed\, the application of nonparametric techniques could be v
ery useful to estimate more accurately the distribution of these statistic
s and consequently those critic values.\n\nFor the validation of the propo
sed algorithm\, several scenarios were created in a simulation study where
the statistics h and k were generated from different distributions such a
s Normal\, Laplace\, and Skew Normal where sample size and the number of l
aboratories are considered. Also\, emphasize on the power of the test to v
erify the capacity of the methodology for detect inconsistencies.\n\nAs ge
neral result\, the new bootstrap methodology presents better results than
those obtained using the parametric traditional methodology\, essentially
when the data is generated by a Skew distribution and the sample size is s
mall. Finally\, this methodology was applied to a real case study of data
obtained through a computational technique of hematic biometry between cli
nical laboratories and a dataset corresponding to serum glucose testing im
plemented on ILS R package.\n\nhttps://conferences.enbis.org/event/11/cont
ributions/228/
LOCATION:Room 5
URL:https://conferences.enbis.org/event/11/contributions/228/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Image-Based Feedback Control Using Tensor Analysis
DTSTART:20210915T133000Z
DTEND:20210915T140000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-203@conferences.enbis.org
DESCRIPTION:Speakers: Jianjun Shi (Georgia Tech)\, Zhen Zhang (Georgia Tec
h)\, Kamran Paynabar (Georgia Tech)\n\nIn manufacturing systems\, many qua
lity measurements are in the form of images\, including overlay measuremen
ts in semiconductor manufacturing\, and dimensional deformation profiles o
f fuselages in an aircraft assembly process. To reduce the process variabi
lity and ensure on-target quality\, process control strategies should be d
eployed\, where the high-dimensional image output is controlled by one or
more input variables. To design an effective control strategy\, one first
needs to estimate the process model off-line by finding the relationship b
etween the image output and inputs\, and then to obtain the control law by
minimizing the control objective function online. The main challenges in
achieving such a control strategy include (i) the high-dimensionality of t
he output in building a regression model\, (ii) the spatial structure of i
mage outputs and the temporal structure of the images sequence\, and (iii)
non-iid noises. To address these challenges\, we propose a novel tensor-b
ased process control approach by incorporating the tensor time series and
regression techniques. Based on the process model\, we can then obtain the
control law by minimizing a control objective function. Although our prop
osed approach is motivated by the 2D image case\, it can be extended to th
e higher-order tensors such as point clouds. Simulation and case studies s
how that our proposed method is more effective than benchmarks in terms of
relative mean square error.\n\nhttps://conferences.enbis.org/event/11/con
tributions/203/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/203/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Accreditation of statisticians
DTSTART:20210915T122000Z
DTEND:20210915T124000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-222@conferences.enbis.org
DESCRIPTION:Speakers: Magnus Pettersson\n\nAccreditation of statisticians
has been offered by ASA and RSS earlier\, and since 2020 by FENStatS. \n\n
The purpose of accreditation is to focus on the professionality\, developm
ent and quality of applied statistical work. We believe that the need for
good statistics and good statisticians is increasing and an accreditation
programme can provide one tool in this process. \n\nThe accreditation summ
arizes the progress and professionality of the applicant. It is a career
path for especially applied statistians that adds value to the universiy e
xam. \n\nAn applicant shall provide proof of:\n\nA - Education\, minimum a
MSc according to Bologna process\nB - Experience\, minimum 5 years work e
xperience\nC - Development\, ongoing professional development\nD - Communi
cation\, samples of work done\nE - Ethics\, knowledge and adherence to rel
evant ethical standards\nF - Membership in a FENStatS member association\n
\nFENStatS provides\, in cooperation with its member organisation\, a stan
dardised system for accreditation that is valid in all its member area. Cu
rrently\, accreditation is availible for by members in Austria\, France\,
Italy\, Portugal\, Spain\, Sweden and Switzerland. \n\nFENStatS accreditat
ion is also mutually recognisied with ASA\, PStat(R). \n\nFurther informat
ion about FENStatS accreditation can be found at: www.fenstats.eu. Applica
tions are submitted through the application portal at the same page.\n\nht
tps://conferences.enbis.org/event/11/contributions/222/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/222/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Vibration signal analysis to classify spur gearbox failure.
DTSTART:20210914T154500Z
DTEND:20210914T160500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-152@conferences.enbis.org
DESCRIPTION:Speakers: Antonio Pérez-Torres (Universidad Politécnica de V
alencia)\, Ana Debón (Universidad Politécnica de Valencia)\, Susana Barc
eló-Cerda (Universidad Politécnica de Valencia)\, René-Vinicio Sánchez
(Universidad Politécnica Salesiana)\n\nA gearbox is a fundamental compon
ent in a rotating machine\; therefore\, detecting a fault or malfunction i
s indispensable early to avoid accidents\, plan maintenance activities and
reduce downtime costs. The vibration signal is widely used to monitor the
condition of a gearbox because it reflects the dynamic behavior in a non-
invasive way. The objective of this research was to perform a ranking of c
ondition indicators to classify the severity level of a series mechanical
faults efficiently. \nThe vibration signal was acquired with six accelero
meters located in different positions by modifying the load and frequency
of rotation using a spur gearbox with different types and severity levels
of failures simulated in laboratory conditions. Firstly\, to summarize the
vibration signal condition\, indicators (statistical parameters)\, both i
n time and frequency domain were calculated. Then\, Random Forest (RF) sel
ected the leading condition indicators\, and finally\, the k nearest neigh
bors and RF ranking methods were used and compared for the severity level.
\nIn conclusion\, the leading condition indicators were determined for th
e time and frequency domain to classify the severity level\, being the mos
t efficient classification method Random Forest.\n\nhttps://conferences.en
bis.org/event/11/contributions/152/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/152/
END:VEVENT
BEGIN:VEVENT
SUMMARY:ShapKit: a Python module dedicated to local explanation of machine
learning models
DTSTART:20210914T144500Z
DTEND:20210914T150500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-138@conferences.enbis.org
DESCRIPTION:Speakers: Vincent Thouvenot\, Simon Grah (OCTO Technology )\n\
nMachine Learning is enjoying an increasing success in many applications:
defense\, cyber security\, etc. However\, models are often very complex. T
his is problematic\, especially for critical systems\, because end-users n
eed to fully understand the decisions of an algorithm (e.g. why an alert h
as been triggered or why a person has a high probability of cancer recurre
nce). One solution is to offer an interpretation for each individual predi
ction based on attribute relevance. Shapley Values\, coming from cooperati
ve game theory\, allow to distribute fairly contributions for each attribu
te in order to understand the difference between a predicted value for an
observation and a base value (e.g. the average prediction of a reference p
opulation). While these values have many advantages\, including their theo
retical guarantees\, they have a strong drawback: the complexity increases
exponentially with the number of features. In this talk\, we will present
and demonstrate ShapKit\, a Python module developed by Thales and availab
le in Open Source dedicated to Shapley Values computation in an efficient
way for local explanation of machine learning\nmodel. We will apply ShapKi
t on a cybersecurity use case.\n\nhttps://conferences.enbis.org/event/11/c
ontributions/138/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/138/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Design-Expert and Stat-Ease360: Easy and Efficient as Illustrated
by Examples
DTSTART:20210915T133000Z
DTEND:20210915T140000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-229@conferences.enbis.org
DESCRIPTION:Speakers: Martin Bezener (StatEase®)\n\nThe book "Application
s of DoE in Engineering and Science" by Leonard Lye contains a wealth of d
esign of experiments (DOE) case studies\, including factorial designs\, fr
actional factorial designs\, various RSM designs\, and combination designs
. A selection of these case studies will be presented using the latest ver
sion of Design Expert®\, a software package developed for use in DOE appl
ications\, and Stat-Ease®360\, a cutting-edge advanced statistical engine
ering package. The presentation includes the design creation as well as th
e analysis of the data. The talk will allow interaction with the attendees
by discussing every step of building the design as well as the analysis o
f the data. This demonstration will prove the ease and the thoroughness of
Stat-Ease software.\n\nReference:\nLye\, L.M. (2019) Applications of DOE
in Engineering and Science: A Collection of 26 Case Studies\, 1st ed.\n\nh
ttps://conferences.enbis.org/event/11/contributions/229/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/229/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Application of domain-specific language models for quality and tec
hnical support in the Food and Beverage Industry
DTSTART:20210914T154500Z
DTEND:20210914T160500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-205@conferences.enbis.org
DESCRIPTION:Speakers: Astrid Kyhl\, Noah Schellenberg\, Alberto Barroso\,
Chiara Mondino\, Peng Liu\n\nIssue Resolution is a critical process in the
manufacturing sector to sustain productivity and quality\, especially in
the Food and Beverage Industry\, where aseptic performance is critical. As
a leader in this industry\, Tetra Pak has built a database regarding qual
ity events reported by Tetra Pak technicians\, each containing domain know
ledge from experts. In this paper\, we present a model framework we have i
nternally developed\, which is using a domain-specific language model to a
ddress two primary natural language challenges impacting the resolution ti
me: \n\n1. Automatically classify a new reported event to the proper exist
ing class \n2. Suggest existing solutions when a new event is being repor
ted\, ranked by relevance of the descriptions of the issues (free text doc
umented by the technician) \n\nOur study shows that the language model cou
ld benefit from training on domain-specific data compared with those train
ed on open-domain data. For task 1\, the language model is trained on the
domain-specific data with an accuracy of over 85%. F1 score average is ove
r 80%. For task 2\, the domain-specific deep learning model is combined w
ith a bag-of-words retrieval function-based algorithm to build an advanced
search engine with an average precision of 53%.\n\nhttps://conferences.en
bis.org/event/11/contributions/205/
LOCATION:Room 5
URL:https://conferences.enbis.org/event/11/contributions/205/
END:VEVENT
BEGIN:VEVENT
SUMMARY:The Shiryaev-Roberts Control Chart for Markovian Count Time Series
DTSTART:20210914T144500Z
DTEND:20210914T150500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-129@conferences.enbis.org
DESCRIPTION:Speakers: Sebastian Ottenstreuer\n\nThe research examines the
zero-state and the steady-state behavior of the Shiryaev-Roberts (SR) proc
edure for Markov-dependent count time series\, using the Poisson INARCH(1)
model as the representative data-generating count process. For the purpos
e of easier evaluation\, the performance is compared to existing CUSUM res
ults from the literature. The comparison shows that SR performs at least a
s well as its more popular competitor in detecting changes in the process
distribution. In terms of usability\, however\, the SR procedure has a pra
ctical advantage\, which is illustrated by an application to a real data s
et. In sum\, the research reveals the SR chart to be the better tool for m
onitoring Markov-dependent counts.\n\nhttps://conferences.enbis.org/event/
11/contributions/129/
LOCATION:Room 5
URL:https://conferences.enbis.org/event/11/contributions/129/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Real-time monitoring of functional data
DTSTART:20210914T152500Z
DTEND:20210914T154500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-188@conferences.enbis.org
DESCRIPTION:Speakers: Fabio Centofanti (University of Naples)\, Antonio Le
pore (Università degli Studi di Napoli Federico II - Dept. of Industrial
Engineering)\, Murat Kulahci (Technical University of Denmark\, Department
of Applied Mathematics and Computer Science\; Luleå University of Techno
logy\, Department of Business Administration\, Technology and Social Scien
ces)\, Max Peter Spooner (Technical University of Denmark\, Department of
Applied Mathematics and Computer Science)\n\nRecent improvements in data a
cquisition technologies have produced data-rich environments in every fiel
d. Particularly relevant is the case where data are apt to be modelled as
functions defined on multidimensional domain\, which are referred to as f
unctional data. A typical problem in industrial applications deals with e
valuating the stability over time of some functional quality characteristi
cs of interest. To this end\, profile monitoring is the suite of statisti
cal process control (SPC) methods that deal with quality characteristics t
hat are functional data. While the main aim of the profile monitoring meth
ods is to assess the stability of the functional quality characteristic\,
in some applications\, the interest relies in understanding if the process
is working properly before its completion\, i.e.\, in the real-time monit
oring of a functional quality characteristic. This work presents a new s
olution to this task\, based on the idea of real-time alignment and simult
aneous monitoring of phase and amplitude variations. The proposal is to it
eratively apply at each time point a procedure consisting of three main st
eps: i) alignment of the partially observed functional data to the referen
ce observation through a registration procedure\; ii) dimensionality reduc
tion through a modification of the functional principal component analysis
(FPCA) specifically designed to consider the phase variability\; iii) mon
itoring of the resulting coefficients. The effectiveness of the proposed m
ethod is demonstrated through both an extensive Monte Carlo simulation and
a real-data example.\n\nhttps://conferences.enbis.org/event/11/contributi
ons/188/
LOCATION:Room 5
URL:https://conferences.enbis.org/event/11/contributions/188/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Hypothesis-based acceptance sampling for modules F and F1 of the E
uropean Measuring Instruments Directive
DTSTART:20210915T130000Z
DTEND:20210915T133000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-130@conferences.enbis.org
DESCRIPTION:Speakers: Cord A. Müller (Deutsche Akademie für Metrologie\,
Bayerisches Landesamt für Maß und Gewicht)\, Clemens Elster (Physikalis
ch-Technische Bundesanstalt)\, Katy Klauenberg (Physikalisch-Technische Bu
ndesanstalt (PTB))\n\nMillions of measuring instruments are verified each
year before being placed on the markets worldwide. In the EU\, such initia
l conformity assessments are regulated by the Measuring Instruments Direct
ive (MID) and its modules F and F1 allow for statistical acceptance sampli
ng. \n\nThis paper re-interprets the acceptance sampling conditions formul
ated by the MID in the formal framework of hypothesis testing. The new int
erpretation is contrasted with the one advanced in WELMEC guide 8.10 [1]\,
and its advantages are elaborated. Besides the conceptual advantage of ag
reeing with a well-known\, statistical framework\, the new interpretation
entails also economic advantages. Namely\, it bounds the producers' risk f
rom above\, such that measuring instruments with sufficient quality are ac
cepted with a guaranteed probability of no less than 95%. Furthermore\, th
e new interpretation applies unambiguously to finite-sized lots\, even ver
y small ones. A new acceptance sampling scheme is derived\, because re-int
erpreting the MID conditions implies that currently available sampling pla
ns are either not admissible or not optimal. \n\nWe conclude that the new
interpretation is to be preferred and suggest re-formulating the statistic
al sampling conditions in the MID. Exchange with WELMEC WG 8 is ongoing to
revise its guide 8.10 and to recommend application of the new sampling sc
heme. \n\n[1] WELMEC European Cooperation in Legal Metrology: Working Grou
p 8 (2018)\, “Measuring Instruments Directive (2014/32/EU): Guide for Ge
nerating Sampling Plans for Statistical Verification According to Annex F
and F1 of MID 2014/32/EU”\n\nhttps://conferences.enbis.org/event/11/cont
ributions/130/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/130/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Statistical analysis of simulation experiments: Challenges for ind
ustrial applications
DTSTART:20210914T123000Z
DTEND:20210914T130000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-177@conferences.enbis.org
DESCRIPTION:Speakers: Bertrand Iooss\n\nThis talk will concern developing
and disseminating statistical tools for answering some industrial issues.
It will be fully based on my 20-years’ experience as a statistician rese
arch engineer and expert in the French research institute of nuclear energ
y (CEA) and the French company of electricity (EDF). I will particularly f
ocus on the domain of uncertainty quantification in numerical simulation a
nd computer experiments modeling. For my company\, in a small-size data co
ntext (that occur in the frequent cases of expensive experiments and/or li
mited available information)\, the numerical model exploration techniques
allow to better understand a risky situation and\, sometimes\, to solve a
safety issue. I will highlight some successful projects (always collective
)\, emphasizing on the scientific innovative parts (kriging metamodeling a
nd global sensitivity analysis in high dimension) but also the organizatio
nal reasons of the success.\n\nhttps://conferences.enbis.org/event/11/cont
ributions/177/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/177/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Detecting changes in Multistream Sequences
DTSTART:20210914T130000Z
DTEND:20210914T133000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-140@conferences.enbis.org
DESCRIPTION:Speakers: George Moustakides (University of Patras)\n\nMultipl
e statistically independent data streams are being observed sequentially a
nd we are interested in detecting\, as soon as possible\, a change in thei
r statistical behavior. We study two different formulations of the change
detection problem. 1) In the first a change appears at a single unknown st
ream but then the change starts switching from one stream to the other fol
lowing a switching mechanism for which we have absolutely no prior knowled
ge. Under the assumption that we can sample simultaneously all streams\, w
e identify the exactly optimum sequential detector when the streams are ho
mogeneous while we develop an asymptotically optimum solution in the inhom
ogeneous case. 2) The second formulation involves a permanent change occur
ring at a single but unknown stream and\, unlike the previous case\, we ar
e allowed to sample only a single stream at a time. We propose a simple de
tection structure based on the classical CUSUM test which we successfully
justify by demonstrating that it enjoys a strong asymptotic optimality pro
perty.\n\nhttps://conferences.enbis.org/event/11/contributions/140/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/140/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Analysis of resistance of spot welding process data in the automot
ive industry via functional clustering techniques
DTSTART:20210914T104000Z
DTEND:20210914T110000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-207@conferences.enbis.org
DESCRIPTION:Speakers: Christian Capezza (University of Naples Federico II)
\, Biagio Palumbo (University of Naples Federico II)\, Antonio Lepore (Uni
versity of Naples Federico II)\, Fabio Centofanti (University of Naples Fe
derico II)\n\nQuality evaluation of resistance spot welding (RSW) joints o
f metal sheets in the automobile sector is generally dependent on expensiv
e and time-consuming offline testing\, which are impracticable in full-sca
le manufacturing on a vast scale. A great opportunity to face this problem
is the increasing digitization in the industry 4.0 framework\, which make
s on-line measurements of process parameters available for every joint man
ufactured. Among possible parameters that can be monitored\, the so-called
dynamic resistance curve (DRC) is considered as the spot welds' technolog
ical signature. This work aims to demonstrate in this context the potentia
l and practical relevance of clustering algorithms to functional data\, i.
e.\, data represented by curves varying over a continuum. The objective is
to partition DRCs into homogenous groups related to spot welds with commo
n mechanical and metallurgical characteristics. The functional data approa
ch has the advantage that it does not need feature extraction\, which is a
rbitrary and problem specific.\nWe discuss the most promising functional c
lustering techniques and apply them to a real-case study on DRCs acquired
during lab tests at Centro Ricerche Fiat. Through the functional clusterin
g approach\, we found that the partitions obtained appear to be related to
the electrodes wear status\, which is surmised to affect the final qualit
y of the RSW joint. R code and the ICOSAF project data are made available
at https://github.com/unina-sfere/funclustRSW/\, where we provide also an
essential tutorial on how to implement the proposed clustering algorithms.
\n\nhttps://conferences.enbis.org/event/11/contributions/207/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/207/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Adhesive bonding process optimization via Gaussian Process models
DTSTART:20210915T102000Z
DTEND:20210915T104000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-141@conferences.enbis.org
DESCRIPTION:Speakers: Ivo Couckuyt (IDLab\, Ghent University - imec)\, Nas
rulloh Loka (IDLab\, Ghent University - imec)\, Bart Van Doninck (Product
ionS\, Flanders Make)\, Alejandro Morales Hernandez (Decision Sciences I
nstitute\, Hasselt University)\, Maarten Witters (ProductionS\, Flanders M
ake)\, Inneke Van Nieuwenhuyse (Research Group Logistics\, Hasselt Unive
rsity)\, Jeroen Jordens (ProductionS\, Flanders Make)\n\nAdhesives are inc
reasingly used in the manufacturing industry because of their desirable ch
aracteristics e.g. high strength-to-weight ratio\, design flexibility\, da
mage tolerance and fatigue resistance. The manufacturing of adhesive joint
s involves a complex\, multi-stage process in which product quality parame
ters\, such as joint strength and failure mode\, are highly impacted by th
e applied process parameters. Optimization of the bonding process paramete
rs is therefore important to guarantee the final product quality and minim
ize production costs.\n\nAdhesive bonding processes are traditionally dete
rmined through expert knowledge and trial and error\, varying only one fac
tor at a time. This approach generally yields suboptimal results and depen
ds highly on the experience and knowledge of the process designer. Additio
nally\, the bonding process parameters\, jointly determine performance and
cost metrics in a complex\, nonlinear way. Therefore\, a more efficient o
ptimization method is desired.\n\nThis research discusses the use of Desig
n of Experiments with Bayesian Optimization and Gaussian process models to
optimize six bonding process parameters for maximal joint strength. The a
pproach was first applied in a simulation environment and later validated
via physical experiments. In the intermediate result\, this novel method s
howed 2% reduction in production cost and 15% reduction in optimal solutio
n search\, compared to the traditional approach with similar joint strengt
hs. Final results will be presented at the conference.\n\nThis research re
ceived funding from the Flemish Government under the “Onderzoeksprogramm
a Artificiële Intelligentie AI Vlaanderen” program. This research was s
upported or partially supported by Flanders Make vzw.\n\nhttps://conferenc
es.enbis.org/event/11/contributions/141/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/141/
END:VEVENT
BEGIN:VEVENT
SUMMARY:An algorithm for robust designs against data loss
DTSTART:20210915T104000Z
DTEND:20210915T110000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-145@conferences.enbis.org
DESCRIPTION:Speakers: Roberto Fontana\, Fabio Rapallo (Università di Geno
va)\n\nOptimal experimental designs are extensively studied in the statist
ical literature. In this work we focus on the notion of robustness of a de
sign\, i.e. the sensitivity of a design to the removal of design points. T
his notion is particularly important when at the end of the experimental a
ctivity the design may be incomplete i.e. response values are not availabl
e for all the points of the design itself. We will see that the definition
of robustness is also related\, but not equivalent\, to D-optimality.\nTh
e methodology for studying robust designs is based on the circuit basis of
the design model matrix. Circuits are minimal dependent sets of the rows
of the design model matrix and provide a representation of its kernel with
special properties. The circuit basis can be computed through several pac
kages for symbolic computation.\nWe present a simple algorithm for finding
robust fractions of a specified size. The basic idea of the algorithm is
to improve a given fraction by exchanging\, for a certain number of times\
, the worst point of the fraction with the best point among those which ar
e in the candidate set but not in the fraction. Some practical examples ar
e presented\, from classical combinatorial designs to two-level factorial
designs including interactions.\n\nhttps://conferences.enbis.org/event/11/
contributions/145/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/145/
END:VEVENT
BEGIN:VEVENT
SUMMARY:The numerical statistical fan and model selection
DTSTART:20210915T104000Z
DTEND:20210915T110000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-131@conferences.enbis.org
DESCRIPTION:Speakers: Sonja Kuhnt (Dortmund University of Applied Sciences
and Arts)\, Arkadius Kalka (Dortmund University of Applied Sciences and A
rts)\n\nIdentifiability of polynomial models is a key requirement for mult
iple\nregression. We consider an analogue of the so-called statistical fan
\, the set of\nall maximal identifiable hierarchical models\, for cases of
noisy design of experiments or measured covariate vectors with a given to
lerance vector. This\ngives rise to the definition of the numerical statis
tical fan. It includes all\nmaximal hierarchical models that avoid approxi
mate linear dependence of the\ndesign vectors. We develop an algorithm to
compute the numerical statistical\nfan using recent results on the computa
tion of all border bases of a design\nideal from the field of algebra. \nI
n the low-dimensional case and for sufficiently small data sets the numeri
cal statistical fan is effectively computable and much smaller than the re
spective statistical fan. The gained\nenhanced knowledge of the space of a
ll stable identifiable hierarchical models\nenables improved model selecti
on procedures. We combine the recursive computation of the numerical stati
stical fan with model selection procedures for linear models and GLMs\, an
d we provide implementations in R.\n\nhttps://conferences.enbis.org/event/
11/contributions/131/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/131/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Greenfield Challenge 2021
DTSTART:20210913T161500Z
DTEND:20210913T171500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-236@conferences.enbis.org
DESCRIPTION:Speakers: Antonella Bodini (CNR-IMATI)\n\nI will present a bri
ef overview of my most recent experiences in disseminating statistical cul
ture: participation in the virtual event STEMintheCity 2020 and the creati
on of statistics pills for a general public\, available on the Outreach we
bsite of the National Resear Council of Italy.\n\nI will conclude with a s
hort presentation of the ongoing multidisciplinary research activity on ca
rdiology and of the related aspects of dissemination.\n\nhttps://conferenc
es.enbis.org/event/11/contributions/236/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/236/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Study of the effectiveness of Bayesian kriging for the decommissio
ning and dismantling of nuclear sites.
DTSTART:20210915T120000Z
DTEND:20210915T122000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-133@conferences.enbis.org
DESCRIPTION:Speakers: Marielle Crozet (CEA)\, Céline Lacaux (Laboratoire
de Mathématiques d'Avignon)\, Bertrand Iooss (EDF R&D)\, Martin Wieskotte
n (CEA)\n\nThe decommissioning of nuclear infrastructures such as power pl
ants arises as these facilities age and come to the end of their lifecycle
. The decommissioning projects expect a complete radiological characteriza
tion of the site\, of both the soil and the civil engineering structure to
optimize efficiency and minimize the costs of said project. To achieve su
ch goal\, statistical tools such as geostatistics are used for the spatial
characterization of radioactive contamination. One of the recurring probl
em using kriging is its sensitivity to parameters estimation. Even though
tools such as the variogram are available for parameter estimation\, they
do not allow for uncertainty quantification in parameter estimation\, lead
ing to over-optimistic prediction variances. A solution to this problem is
Bayesian kriging\, which takes into account uncertainty in parameter esti
mation by considering parameters as random variables and assigning them pr
ior specifications. We chose to study the efficiency of Bayesian kriging i
n comparison with standard kriging methods\, by varying the size of the da
ta set available\, and tested its effectiveness against misspecification\,
such as wrong priors hyperparameters or covariance models. These comparis
ons were made on simulated data sets\, as well as on a real data set from
the decommissioning project of the G3 reactor in CEA Marcoule.\n\nhttps://
conferences.enbis.org/event/11/contributions/133/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/133/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Customer prioritization for marketing actions
DTSTART:20210915T104000Z
DTEND:20210915T110000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-153@conferences.enbis.org
DESCRIPTION:Speakers: Ignasi Puig\, Xavier Puig\, Daniel González\n\nSele
cting customers for marketing actions is an important decision for compani
es. The profitability of a customer and his inactivity risk are two import
ant aspects of this selection process. These indicators can be obtained us
ing the known Pareto/NBD model. This work proposes clustering customers ba
sed on their purchase frequency and purchase value per period before imple
menting the Pareto/NBD model onto each cluster. This initial cluster model
allows estimating the customers purchase value and improves the parameter
estimation accuracy of the Pareto/NBD by using alike individuals in the f
itting. Models are implemented using Bayesian inference as to determine th
e uncertainty behind the different estimates. Finally\, using the outputs
of both models\, the initial cluster and the Pareto/NBD\, the project deve
loped a guideline to classify clients into interpretable groups to facilit
ate their prioritization for marketing actions. The methodology was develo
ped and implemented on a set of 25\,600 sales from a database of 1\,500 cu
stomers from beauty products wholesaler.\n\nhttps://conferences.enbis.org/
event/11/contributions/153/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/153/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Commented Summary of a Year of Work in Covid-19 Statistical Modeli
ng
DTSTART:20210915T122000Z
DTEND:20210915T124000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-123@conferences.enbis.org
DESCRIPTION:Speakers: Jorge Romeu (Emeritus State Univ. of NY (SUNY))\n\nW
e summarize eleven months of pro-bono work on statistical modeling and ana
lysis of Covid-19 topics. For each of the papers and tutorials included he
re we provide a one-paragraph summary and commentary\, including methods u
sed\, results\, and possible public health applications\, as well as the R
esearchGate url to access them. Section 1 is an Introduction. In Section 2
we describe the web page created\, and its main sections. In Section 3 we
summarize three papers on Design of Experiments and Quality Control Appli
cations. In Section 4\, we summarize four papers on Reliability\, Survival
Analysis and Logistics Applications to Vaccine development. In Section 5
we summarize three papers on Multivariate Analysis (Principal Components\,
Discriminant Analyses) and Logistics Regression. In Section 6 we summariz
e three Stochastic Process papers that implement Markov Chain models to an
alyze herd immunization. In Section 7\, we summarize three papers on Socio
-economic analyses of vaccine rollout\, and race\, ethnicity and class pro
blems\, derived from Covid-19. In Section 8\, we conclude\, discussing the
procedures used to produce these papers\, and the audiences we hope to re
ach.\n\nhttps://conferences.enbis.org/event/11/contributions/123/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/123/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A Multivariate Non Parametric Monitoring Procedure Based on Convex
Hulls
DTSTART:20210914T133000Z
DTEND:20210914T140000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-230@conferences.enbis.org
DESCRIPTION:Speakers: Sotiris Bersimis (University of Piraeus\, Greece)\,
Polychronis Economou (University of Patras\, Greece)\, Subha Chakraborti (
University of Alabama\, USA)\n\nBersimis et al. (2007) motivated by Woodal
l and Montgomery (1999) statement published an extensive review paper of t
he field of MSPM. According to Bersimis et al. (2007) open problems in the
field of MSPM\, among others are robust design of monitoring procedures a
nd non-parametric control charts. In this work\, we introduce a non-parame
tric control scheme based on convex hulls. The proposed non-parametric con
trol chart is using bootstrap for estimating the kernel of the multivariat
e distribution and then appropriate statistics based on convex hull are mo
nitored. The performance of the proposed control chart is very promising.\
n\nReferences:\nBersimis\, S.\, Psarakis\, S. and Panaretos\, J. (2007). "
Multivariate statistical process control charts: an overview". Quality and
Reliability Engineering International\, 23\, 517-543.\nWoodall\, W. H. an
d Montgomery\, D. C. (1999). "Research Issues and Ideas in Statistical Pro
cess Control". Journal of Quality Technology\, 31\, 376-386.\n\nhttps://co
nferences.enbis.org/event/11/contributions/230/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/230/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A fixed-sequence approach for selecting best performing classifier
s
DTSTART:20210915T124000Z
DTEND:20210915T130000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-227@conferences.enbis.org
DESCRIPTION:Speakers: Armando Ciardiello (Dept. of industrial Engineering)
\, Amalia Vanacore (Department of Industrial Engineering University of Nap
les Federico II)\, Maria Sole Pellegrino (Dept. of Industrial Engineering)
\n\nAn important issue in classification problems is the comparison of cla
ssifiers predictive performance\, commonly measured as proportion of corre
ct classifications and often referred to as accuracy or similarity measure
. \nThis paper suggests a two-step fixed-sequence approach in order to ide
ntify the best performing classifiers among those selected as suitable for
the problem at hand. At the first step of the fixed-sequence approach\, t
he hypothesis that each classifier accuracy exceeds a desired performance
threshold is tested via a simultaneous inference procedure accounting for
the joint distribution of individual test statistics and the correlation b
etween them. At the second step\, focusing only on classifiers selected at
first step\, significant performance differences are investigated via a h
omogeneity test. \nThe applicability and usefulness of the two-step approa
ch is illustrated through two real case studies concerning nominal and ord
inal multi-class classification problems. The accuracy of three machine le
arning algorithms (i.e. Deep Neural Network\, Random Forest\, Extreme Grad
ient Boosting) is assessed via Gwet’s Agreement Coefficient (AC) and com
pared against similarity measure and Cohen Kappa. Case studies results rev
eal the absence of paradoxical behavior in AC coefficient and the positive
effect of a weighting scheme accounting for misclassification severity wi
th ordinal classifications\, shedding light on the advantages of AC as mea
sure of classifier accuracy.\n\nhttps://conferences.enbis.org/event/11/con
tributions/227/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/227/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Forecasting count time series in retail
DTSTART:20210914T092000Z
DTEND:20210914T094000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-208@conferences.enbis.org
DESCRIPTION:Speakers: Bruno Flores (ICMAT-CSIC)\n\nLarge-scale dynamic for
ecasting of non-negative count series is a major challenge in many areas l
ike epidemic monitoring or retail management. We propose Bayesian state-sp
ace models that are flexible enough to adequately forecast high and low co
unt series and exploit cross-series relationships with a multivariate appr
oach. This is illustrated with a large scale sales forecasting problem fac
ed by a major retail company\, integrated within its inventory management
planning methodology. The company has hundreds of shops in several countri
es\, each one with thousands of references.\n\nhttps://conferences.enbis.o
rg/event/11/contributions/208/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/208/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Sparse and smooth cluster analysis of functional data
DTSTART:20210913T141500Z
DTEND:20210913T144500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-189@conferences.enbis.org
DESCRIPTION:Speakers: Biagio Palumbo (University of Naples Federico II)\,
Antonio Lepore (Università degli Studi di Napoli Federico II - Dept. of I
ndustrial Engineering)\, Fabio Centofanti (University of Naples)\n\nThe sp
arse and smooth clustering (SaS-Funclust) method proposed in [1] is presen
ted. The aim is to cluster functional data while jointly detecting the mos
t informative portion(s) of the functional data domain. The SaS-Funclust m
ethod relies on a general functional Gaussian mixture model with parameter
s estimated by maximizing the sum of a log-likelihood function penalized b
y a functional adaptive pairwise penalty and a roughness penalty. The func
tional adaptive penalty is introduced to automatically identify the inform
ative portion of domain by shrinking the means of separated clusters to so
me common values. At the same time\, the roughness penalty imposes some sm
oothness to the estimated cluster means. The proposed method is shown to e
ffectively enhance the solution interpretability while still maintaining f
lexibility in terms of clustering performance. The methods are implemented
and archived in an R package *sasfunclust*\, available on CRAN [2].\n\n[1
] Centofanti\, F.\, Lepore\, A.\, Palumbo\, B. (2021). Sparse and Smooth F
unctional Data Clustering. Preprint arXiv:2103.15224\n[2] Centofanti F.\,
Lepore A.\, Palumbo B. (2021). sasfunclust: Sparse and Smooth Functional C
lustering. R package version 1.0.0. [https://CRAN.R–project.org/package=
sasfunclust]\n\nhttps://conferences.enbis.org/event/11/contributions/189/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/189/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Analyzing categorical time series in the presence of missing obser
vations
DTSTART:20210915T120000Z
DTEND:20210915T122000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-120@conferences.enbis.org
DESCRIPTION:Speakers: Christian Weiß (Helmut Schmidt University)\n\nIn re
al applications\, time series often exhibit missing observations such that
standard analytical tools cannot be applied. While there are approaches o
f how to handle missing data in quantitative time series\, the case of cat
egorical time series seems not to have been treated so far. Both for the c
ase of ordinal and nominal time series\, solutions are developed that allo
w to analyze their marginal and serial properties in the presence of missi
ng observations. This is achieved by adapting the concept of amplitude mod
ulation\, which allows to obtain closed-form asymptotic expressions for th
e derived statistics' distribution (assuming that missingness happens inde
pendently of the actual process). The proposed methods are investigated wi
th simulations\, and they are applied in a project on migraine patients\,
where the monitored qualitative time series on features such as pain peak
severity or perceived stress are often incomplete.\n\nThe talk relies on t
he open-access publication\n\nWeiß (2021) Analyzing categorical time seri
es in the presence of missing observations.\nStatistics in Medicine\, in p
ress.\nhttps://doi.org/10.1002/sim.9089\n\nhttps://conferences.enbis.org/e
vent/11/contributions/120/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/120/
END:VEVENT
BEGIN:VEVENT
SUMMARY:AdaPipe: A Recommender System for Adaptive Computation Pipelines i
n Cyber-Manufacturing Computation Services
DTSTART:20210915T140000Z
DTEND:20210915T143000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-209@conferences.enbis.org
DESCRIPTION:Speakers: Ran Jin\, Xiaoyu Chen (Virginia Tech)\n\nThe industr
ial cyber-physical systems (ICPS) will accelerate the transformation of of
fline data-driven modeling to fast computation services\, such as computat
ion pipelines for prediction\, monitoring\, prognosis\, diagnosis\, and co
ntrol in factories. However\, it is computationally intensive to adapt com
putation pipelines to heterogeneous contexts in ICPS in manufacturing.\nIn
this paper\, we propose to rank and select the best computation pipelines
to match contexts and formulate the problem as a recommendation problem.
The proposed method Adaptive computation Pipelines (AdaPipe) considers sim
ilarities of computation pipelines from word embedding\, and features of c
ontexts. Thus\, without exploring all computation pipelines extensively in
a trial-and-error manner\, AdaPipe efficiently identifies top-ranked comp
utation pipelines. We validated the proposed method with 60 bootstrapped d
ata sets from three real manufacturing processes: thermal spray coating\,
printed electronics\, and additive manufacturing. The results indicate tha
t the proposed recommendation method outperforms traditional matrix comple
tion\, tensor regression methods\, and a state-of-the-art personalized rec
ommendation model.\n\nhttps://conferences.enbis.org/event/11/contributions
/209/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/209/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A Predictive Maintenance Model Proposal for a Manufacturing Compan
y
DTSTART:20210914T104000Z
DTEND:20210914T110000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-224@conferences.enbis.org
DESCRIPTION:Speakers: Cemal Aydın (TÜBİTAK)\, Volkan Sonmez (Hacettepe
University)\n\nMaintenance planning is one of the most important problems
for manufacturing enterprises. Maintenance strategies applied in an indust
ry are corrective and preventive maintenance strategies. The development o
f sensor technologies has led to a widespread use of preventive maintenanc
e methods. However\, it can be costly for small and medium-sized enterpris
es to install such sensor systems. This study aims to propose a predictive
maintenance model based on the loss data of production lines without such
recorded data for production equipment. \nIn the study\, data belonging t
o a company that produces PVC profiles\, such as amount of loss based on s
hift and line\, production speed differences and number of shifts passed o
ver the last maintenance\, were used. At first\, a threshold value was det
ermined considering planned maintenances. Then\, models that estimate the
amount of loss for the production line for the following shift\, were trai
ned. Statistical learning algorithms such as linear regression\, neural ne
tworks\, random forest\, and gradient boosting were used to train the mode
ls. When the performance of the trained models was compared\, it was seen
that the most successful model was the neural network. \nAt the end of the
study\, it is explained how to decide whether to perform maintenance or n
ot for a production line. According to the proposed method\, amount of los
s in the related production line will be estimated and this is compared wi
th the threshold value. If the estimated loss is greater than the threshol
d value\, maintenance should be performed\, otherwise\, no maintenance wil
l be performed.\n\nhttps://conferences.enbis.org/event/11/contributions/22
4/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/224/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Univariate Self-Starting Shiryaev (U3S): A Bayesian Online Change
Point Model for Short Runs
DTSTART:20210914T104000Z
DTEND:20210914T110000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-204@conferences.enbis.org
DESCRIPTION:Speakers: Konstantinos Bourazas (Athens University of Economic
s and Business)\, Panagiotis Tsiamyrtzis (Politecnico di Milano)\n\nIn Sta
tistical Process Control/Monitoring (SPC/M) our interest is in detecting w
hen a process deteriorates from its “in control” state\, typically est
ablished after a long phase I exercise. Detecting shifts in short horizon
data of a process with unknown parameters\, (i.e. without a phase I calibr
ation) is quite challenging. \nIn this work\, we propose a self-starting B
ayesian change point scheme\, which is based on the cumulative posterior p
robability that a change point has been occurred. We will focus our attent
ion on univariate Normal data\, aiming to detect persistent shifts for the
mean or the variance. The proposed methodology is a generalization of Shi
ryaev’s process\, as it allows both the parameters and shift magnitude t
o be unknown. Furthermore\, the Shiryaev’s assumption that the prior pro
bability on the location of the change point is constant will be relaxed.
Posterior inference for the unknown parameters and the location of a (pote
ntial) change point will be provided. \nTwo real data sets will illustrate
the Bayesian self-starting Shiryaev’s scheme\, while a simulation study
will evaluate its performance against standard competitors in the cases o
f mean changes and variance inflations.\n\nhttps://conferences.enbis.org/e
vent/11/contributions/204/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/204/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Harnessing the recondite role of randomization in today's scientif
ic\, engineering\, and industrial world
DTSTART:20210915T144500Z
DTEND:20210915T154500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-231@conferences.enbis.org
DESCRIPTION:Speakers: Tirthankar Dasgupta (Rutgers University)\n\nRandomiz
ed experiment is a quintessential methodology in science\, engineering\, b
usiness and industry for assessing causal effects of interventions on outc
omes. Randomization tests\, conceived by Fisher\, are useful tools to anal
yze data obtained from such experiments because they assess the statistica
l significance of estimated treatment effects without making any assumptio
ns about the underlying distribution of the data. Other attractive feature
s of randomization tests include flexibility in the choice of test statist
ic and adaptability to experiments with complex randomization schemes and
non-standard (e.g.\, ordinal) data. In the past\, these tests' major drawb
ack was their possibly prohibitive computational requirements. Modern comp
uting resources make randomization tests pragmatic\, useful tools driven p
rimarily by intuition. In this talk we will discuss a principled approach
to conducting randomization-based inference in a wide array of industrial
and engineering settings and demonstrate their advantage using examples. W
e will also briefly argue that randomization tests are natural and effecti
ve tools for data fusion\, that is\, combining results from an ensemble of
similar or dissimilar experiments. Finally\, if time permits\, we will al
so discuss how this knowledge can be easily communicated to students and p
ractitioners and mention some available computing resources.\n\nhttps://co
nferences.enbis.org/event/11/contributions/231/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/231/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Hands-on Projects for Teaching DoE
DTSTART:20210914T144500Z
DTEND:20210914T161500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-225@conferences.enbis.org
DESCRIPTION:Speakers: Sonja Kuhnt (Dortmund University of Applied Sciences
and Arts)\, Shirley Coleman\n\n__About the Session:__\n\nAre you interest
ed in case studies and real-world problems for active learning of statisti
cs? Then come and join us in this one-hour interactive session organised b
y the SIG Statistics in Practice. The session follows on from a similar ev
ent in ENBIS 2020. \nA famous project for students to apply the acquired k
nowledge of design of experiments is Box's paper helicopter. Although bein
g quite simple and cheap to build\, it covers various aspects of DoE. Beyo
nd this\, what other possible DoE projects are realistic in a teaching en
vironment? What are your experiences in using them? Can we think of new on
es? There are lots of ideas we could explore\, involving more complex scen
arios like time series dependents with cross overs\, functional data analy
sis\, as well as mixture experiments.\nWe want to share projects\, discuss
pitfalls and successes and search our mind for new ideas. Come and join u
s for this session. You may just listen\, enjoy and hopefully contribute t
o the discussion or even share a project idea. \n\n\n__Planned Contributio
ns:__\n\nNadja Bauer (SMF and Dortmund University of Applied Sciences and
Arts\, Germany) presents a __color mixing DoE problem__\, where the adjust
able parameters such as\, among others\, the proportion and temperature of
the incoming colors (cyan\, magenta and yellow) influence the color and t
emperature of the mixture.\n\nMark Anderson\, lead author of the DOE/RSM/F
ormulation Simplified book trilogy\, will demonstrate a fun __experiment o
n bouncing balls__ that illustrates the magic of multifactor DoE.\n\nJacqu
eline Asscher (Kinneret College on the Sea of Galilee and Technion\, Israe
l) shares her __water beads DoE project__. Water beads are small\, cheap b
alls made from a water-absorbing polymer. They are added to the soil in ga
rdens and planters\, as they absorb a large amount of water and release it
slowly. This is a simple but not entirely trivial process. It can be inve
stigated using experiments run either at home or in the classroom.\n\nJona
than Smyth-Renshaw (Jonathan Smyth-Renshaw & Associates Limited\, UK) pres
ents a __DoE problem with a food manufacturer__\, where a Plackett and Bur
man Design experiment is used to understand the impact of 7 factors - 5 in
gredients and 2 process settings. \n\nThejasvi TV (India) presents applica
tions of __DoE in dentistry__.\n\nhttps://conferences.enbis.org/event/11/c
ontributions/225/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/225/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Outliers and the instrumental variables estimator in the linear re
gression model with endogeneity
DTSTART:20210915T122000Z
DTEND:20210915T124000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-148@conferences.enbis.org
DESCRIPTION:Speakers: Aleš Toman (School of Economics and Business\, Univ
ersity of Ljubljana)\n\nIn a linear regression model\, endogeneity (i.e.\,
a correlation between some explanatory variables and the error term) make
s the classical OLS estimator biased and inconsistent. When instrumental v
ariables (i.e.\, variables that are correlated with the endogenous explana
tory variables but not with the error term) are available to partial out e
ndogeneity\, the IV estimator is consistent and widely used in practice. T
he effect of outliers on the OLS estimator is carefully studied in robust
statistics\, but surprisingly\, the effect of outliers on the IV estimator
has received little attention in previous research\, with existing work m
ostly focusing on robust covariance estimation.\n\nIn this presentation\,
we use the forward search algorithm to investigate the effect of outliers
(and other contamination schemes) on various aspects of the IV-based estim
ation process. The algorithm begins the analysis with a subset of observat
ions that does not contain outliers and then increases the subset by addin
g one observation at a time until all observations are included and the en
tire sample is analyzed. Contaminated observations are included in the sub
set in the final iterations. During the process\, various statistics and r
esiduals are monitored to detect the effects of outliers. \n\nWe use simul
ation studies to investigate the effect of known outliers occurring in the
(i) dependent\, (ii) exogenous or (iii) endogenous exploratory\, or (iv)
instrumental variable. Summarizing the results\, we propose and implement
a method to identify outliers in a real data set where contamination is no
t known in advance.\n\nhttps://conferences.enbis.org/event/11/contribution
s/148/
LOCATION:Room 5
URL:https://conferences.enbis.org/event/11/contributions/148/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Explainable AI and Predictive Maintenance
DTSTART:20210914T140000Z
DTEND:20210914T143000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-226@conferences.enbis.org
DESCRIPTION:Speakers: Florian Sobieczky (Software Competence Center Hagenb
erg GmbH)\, Michael Roßbory (Software Competence Center Hagenberg GmbH)\n
\nNon-linear predictive machine learning models (such as deep learning) ha
ve emerged as a successful approach in many industrial applications\, as t
he accuracy of predictions often surpasses classical statistical approache
s in a significant\, and also effective way. Predictive maintenance tasks
(such as predicting change points or detecting anomalies) are particularly
susceptible to this improvement. However\, the ability to interpret the i
ncrease in accuracy isn't generally delivered alongside with the applicati
on of these models. In several manufacturing scenarios\, however\, a presc
riptive solution is in high demand. The talk surveys several methods to re
nder non-linear predictive models for time series data explainable and als
o introduces a new change point detection technique involving a Long Short
Term Memory neural network. The focus on time series is due to the specif
ic need of methods for this data type in manufacturing and therefore predi
ctive maintenance scenarios.\n\nhttps://conferences.enbis.org/event/11/con
tributions/226/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/226/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Constructing nonparametric control charts for correlated and indep
endent data using resampling techniques
DTSTART:20210914T150500Z
DTEND:20210914T152500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-180@conferences.enbis.org
DESCRIPTION:Speakers: Priscila Guayasamín (Dep. de Matemática\, Escuela
Politécnica Nacional)\, Javier Tarrío-Saavedra (MODES\, CITIC\, Universi
dade da Coruña\, Escola Politécnica Superior)\, Salvador Naya (MODES\, C
ITIC\, ITMATI\, Universidade da Coruña\, Escola Politécnica Superior)\,
Rubén Fernández-Casal (Dep. de Matemáticas\, Universidade da Coruña\,
Spain)\, Miguel Flores (MODES\,SIGTIG\, Dep. de Matemática\, Escuela Pol
itécnica Nacional)\n\nNon-parametric control charts based on data depth a
nd resampling techniques are designed to monitor multivariate independent
and dependent data.\n\nPhase I\n-------\n\nDependent and independent case\
n\n 1. The depths $ D_F (X_i) $ ordered in ascending order are obtained.\n
2. The lower control limit $ (LCI) $ is calculated as the quantile at the
$ \\alpha $ level of the observations under null hypothesis such that the
percentage of false alarms are approximately equal to $ \\alpha $.\n 3. I
f $ D (X_i) \\leq LCI $ then the process is out of control.\n\nFor the est
imation of the quantile\, smoothing bootstrap\, stationary bootstrap have
been applied for independent and dependent case.\n\nPhase II\n--------\n\n
1. From the reference sample $ \\{X_1\, ...\, X_n \\} $ the depth of the
data $ D(X_i) $ is calculated with $ i = 1\, ...\, n $ and based on this t
he depths of the monitoring sample $ D(Y_j) $ are obtained with $ j = n +
1\, ...\, m $ based on the calibration sample\n 2. Monitor the process\, i
f you have observations $ D (Y_j) \\leq LCL $ then the process is out of c
ontrol.\n 3. Calculate the percentage of rejection as the average of obser
vations under the lower control limit.\n\nThe simplicial depth in general
has a better performance for all sample sizes. It is noted that as the sam
ple size increases\, the Tukey and Simplicial measures yield better result
s.\n\nhttps://conferences.enbis.org/event/11/contributions/180/
LOCATION:Room 5
URL:https://conferences.enbis.org/event/11/contributions/180/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Design Optimization for the Step-Stress Accelerated Degradation Te
st under Tweedie Exponential Dispersion Process
DTSTART:20210914T144500Z
DTEND:20210914T150500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-155@conferences.enbis.org
DESCRIPTION:Speakers: David Han\n\nThe accelerated degradation test (ADT)
is a popular tool for assessing the reliability characteristics of highly
reliable products. Henceforth\, designing an efficient ADT has been of gre
at interest\, and it has been studied under various well-known stochastic
degradation processes\, including Wiener process\, gamma process\, and inv
erse Gaussian process. In this work\, Tweedie exponential dispersion proce
ss is considered as a unified model for general degradation paths\, includ
ing the aforementioned processes as special cases. Its flexibility can pro
vide better fits to the degradation data and thereby improve the reliabili
ty analyses. For computational tractability\, the saddle-point approximati
on method is applied to approximate its density. Based on this framework\,
the design optimization for the step-stress ADT is formulated under the C
-optimality. Under the constraint that the total experimental cost does no
t exceed a pre-specified budget\, the optimal design parameters such as me
asurement frequency and test termination time are determined via minimizin
g the approximate variance of the estimated mean time to failure of a prod
uct/device under the normal operating condition.\n\nhttps://conferences.en
bis.org/event/11/contributions/155/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/155/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Inference for the Progressively Type-I Censored $K$-Level Step-Str
ess Accelerated Life Tests Under Interval Monitoring with the Lifetimes fr
om a Log-Location-Scale Family
DTSTART:20210914T150500Z
DTEND:20210914T152500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-157@conferences.enbis.org
DESCRIPTION:Speakers: Aruni Jayathilaka (The University of Texas at San An
tonio)\, David Han\n\nAs the field of reliability engineering continues to
grow and adapt with time\, accelerated life tests (ALT) have progressed f
rom luxury to necessity. ALT subjects test units to higher stress levels t
han normal conditions\, thereby generating more failure data in a shorter
time period. In this work\, we study a progressively Type-I censored k-lev
el step-stress ALT under interval monitoring. In practice\, the financial
and technical barriers to ascertaining precise failure times of test units
could be insurmountable\, therefore\, it is often practical to collect fa
ilure counts at specific points in time during ALT. Here\, the latent fail
ure times are assumed to have a log-location-scale distribution as the obs
erved lifetimes may follow Weibull or log-normal distributions\, which are
members of the log-location-scale family. Here\, we develop the inferenti
al methods for the step-stress ALT under the general log-location-scale fa
mily\, assuming that the location parameter is linearly linked to the stre
ss level. The methods are illustrated using three popular lifetime distrib
utions: Weibull\, lognormal and log-logistic.\n\nhttps://conferences.enbis
.org/event/11/contributions/157/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/157/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Adaptive Design and Inference for a Step-Stress Accelerated Life T
est
DTSTART:20210914T152500Z
DTEND:20210914T154500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-156@conferences.enbis.org
DESCRIPTION:Speakers: Haifa Ismail-Aldayeh (The University of Texas at San
Antonio)\, David Han\n\nAdvancement in manufacturing has significantly ex
tended the lifetime of a product while at the same time it made harder to
perform life testing at the normal operating condition due to the extensiv
ely long life spans. Accelerated life tests (ALT) can mitigate this issue
by testing units at higher stress levels so that the lifetime information
can be acquired more quickly. The lifetime of a product at normal operatio
n can then be estimated through extrapolation using a regression model. Ho
wever\, there are potential technical difficulties since the units are sub
jected to higher stress levels than normal. In this work\, we develop an a
daptive design of a step-stress ALT in which stress levels are determined
sequentially based on the information obtained from the preceding steps. A
fter each stress level\, the estimates of the model parameters are updated
and the decision is made on the direction of the next stress level by usi
ng a design criteria such as D- and C-optimality. Assuming the popular log
-linear assumption between the mean lifetime and stress levels\, this adap
tive design and inference are illustrated based on exponential lifetimes w
ith progressive Type-I censoring.\n\nhttps://conferences.enbis.org/event/1
1/contributions/156/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/156/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Long short-term memory neural network for statistical process cont
rol of autocorrelated multiple stream process with an application to HVAC
systems in passenger rail vehicles
DTSTART:20210914T094000Z
DTEND:20210914T100000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-206@conferences.enbis.org
DESCRIPTION:Speakers: Gianluca Sposito (Department of Industrial Engineeri
ng\, University of Naples Federico II)\, Antonio Lepore (Department of Ind
ustrial Engineering\, University of Naples Federico II)\, Biagio Palumbo (
Department of Industrial Engineering\, University of Naples Federico II)\,
Giuseppe Giannini (Head of Operation Service and Maintenance Product Evol
ution\, Hitachi Rail Group)\n\nRail transport demand in Europe has increas
ed over the last few years\, and passenger thermal comfort has been playin
g a key role in the fierce competition among different transportation comp
anies. Furthermore\, European standards settle operational requirements of
passenger rail coaches in terms of air quality and comfort level. To meet
these standards and the increasing passenger thermal comfort demand\, dat
a from on-board heating\, ventilation and air conditioning (HVAC) systems
have been collected by railway companies to improve maintenance programs i
n the industry 4.0 scenario. Usually\, a train consists of several coaches
equipped with a dedicated HVAC system\, and the sensor signals coming fro
m each HVAC system produce multiple data streams. This setting can thus be
regarded as a multiple stream process (MSP). Unfortunately\, the massive
amounts of data collected at high rates makes each stream more likely to b
e autocorrelated. This scenario calls for a new methodology capable of ove
rcoming the simplifying assumptions on which traditional MSP models are ba
sed. This work is intended to propose a new control charting procedure bas
ed on a long short-term memory neural network trained to solve the binary
classification problem of detecting whether the MSP is in control or out o
f control\, i.e.\, to recognize mean shifts in autocorrelated MSPs. A simu
lation study is performed to assess the performance of the proposed approa
ch and its practical applicability is illustrated by an application to the
monitoring of HVAC system data\, made available by the rail transport com
pany Hitachi Rail based in Italy.\n\nhttps://conferences.enbis.org/event/1
1/contributions/206/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/206/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Variable importance analysis of railway vehicle responses
DTSTART:20210914T133000Z
DTEND:20210914T140000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-158@conferences.enbis.org
DESCRIPTION:Speakers: Bernd Luber (Virtual Vehicle Research GmbH)\, Josef
Fuchs\, Florian Semrad (Siemens Mobility GmbH Österreich)\, Anna Pichler
(Virtual Vehicle Research GmbH)\n\nIn the development process of railway v
ehicles several requirements considering reliability and safety have to be
met. These requirements are commonly assessed by using Multi-Body-Dynamic
s (MBD) simulations and on-track measurements.\nIn general\, the vehicle/t
rack interaction is significantly influenced by varying\, unknown or non-q
uantifiable operating conditions (e.g. coefficient of friction) resulting
in a high variance of the vehicle responses (forces and accelerations). Th
e question is\, which statistical methods allow to identify the significan
t operating conditions to be considered in the simulation?\n\nThis paper p
roposes a methodology to quantify the effects of operating conditions (ind
ependent variables) on vehicle responses (dependent variables) based on me
asurements and simulations. A variable importance analysis is performed co
nsidering the nonlinear behaviour of the vehicle/track interaction as well
as the correlation between the independent variables. Hence\, two statist
ical modelling approaches are considered. The focus is on linear regressio
n models\, which make it possible to include the correlation behaviour of
the independent variables in the analyses. Further\, random forest models
are used to reflect the non-linearity of the vehicle/track interaction.\n\
nThe variable importance measures\, derived from both approaches\, result
in an overview of the effects of operating conditions on vehicle responses
\, considering the complexity of the data. Finally\, the proposed methodol
ogy provides a determined set of operating conditions to be considered in
the simulation.\n\nhttps://conferences.enbis.org/event/11/contributions/15
8/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/158/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Railway track degradation prediction using Wiener process modellin
g
DTSTART:20210914T092000Z
DTEND:20210914T094000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-159@conferences.enbis.org
DESCRIPTION:Speakers: mahdieh sedghi\, Bjarne Bergquist (Luleå University
of Technology)\n\nTrack geometry is critical for railway infrastructures\
, and the geometry condition and the expected degradation rate are vital f
or planning maintenance actions to assure the tracks’ reliability and sa
fety. The degradation prediction accuracy is\, therefore\, essential. The
Wiener process has been widely used for degradation analysis in various ap
plications based on degradation measurements. In railway infrastructure\,
however\, Wiener process-based degradation models are uncommon. This prese
ntation explores the Wiener process for predicting railway track degradati
on. First\, we review different data-driven approaches found in the litera
ture to estimate the Wiener process parameters and updating them when new
measurements are collected. We study different procedures to estimate and
update the Wiener process parameters and evaluate their computational perf
ormance and prediction errors based on measurement data for a track line i
n northern Sweden. The result can help to balance the computational comple
xity and the prediction accuracy when selecting a Wiener process-based deg
radation model for predictive maintenance of the railway track.\n\nhttps:/
/conferences.enbis.org/event/11/contributions/159/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/159/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Six-Sigma and Obesity – Part 1
DTSTART:20210914T102000Z
DTEND:20210914T104000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-165@conferences.enbis.org
DESCRIPTION:Speakers: Roland Caulcutt (Caulcutt Associates)\n\nWhen the Co
vid19 pandemic is no longer the prime burden on British health services\,
it might be possible to refocus on the three concerns that threatened to o
verwhelm the National Health Service in 2019. Namely\, heart disease\, can
cer and obesity.\nWhilst the NHS can reasonably claim to have made progres
s with the first two\, it is faced with an ever-increasing level in obesit
y. To non-clinical members of society this may seem rather surprising\, c
onsidering the relative simplicity of the fat producing process\, compared
with the extreme complexity of cancer and heart disease. It may seem eve
n more surprising to the many statisticians and process improvement profes
sionals who witnessed the great success of blackbelts improving organisati
onal processes whilst working within a culture of “Six-Sigma”.\nPart 1
of this presentation will explain why many blackbelts have had such amazi
ng success by improving organisational processes many of which had a histo
ry of chronic under-performance\n\nhttps://conferences.enbis.org/event/11/
contributions/165/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/165/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Evaluating and Monitoring the Quality of Online Products and Servi
ces via User-Generated Reviews
DTSTART:20210913T134500Z
DTEND:20210913T141500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-171@conferences.enbis.org
DESCRIPTION:Speakers: Qiao Liang\, Kaibo Wang\n\nUser-generated content in
cluding both review texts and user ratings provides important information
regarding the customer-perceived quality of online products and services.
The quality improvement of online products as well as services will benefi
t from a general framework of analyzing and monitoring these user-generate
d content. This study proposes a modeling and monitoring method for online
user-generated content. A unified generative model is constructed to comb
ine words and ratings in customer reviews based on their latent sentiment
and topic assignments\, and a two-chart scheme is proposed for detecting s
hifts of customer responses in dimensions of sentiments and topics\, respe
ctively. The proposed method shows superior performance in shift detection
\, especially for the sentiment shifts in customer responses\, based on th
e results of simulation and a case study.\n\nhttps://conferences.enbis.org
/event/11/contributions/171/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/171/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Classification of On-Road Routes for the Reliability Assessment of
Drive-Assist Systems in Heavy-Duty Trucks based on Electronic Map Data
DTSTART:20210914T130000Z
DTEND:20210914T133000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-163@conferences.enbis.org
DESCRIPTION:Speakers: Nikolaus Haselgruber (CIS consulting in industrial s
tatistics GmbH)\, Boris-Michael Guhr (Daimler Trucks AG)\, Harald Ihle (Da
imler Trucks AG)\n\nThe development of drive assist systems\, such as traf
fic sign recognition and distance regulation\, is one of the most importan
t tasks on the way to autonomous driving. With focus on the definition of
reliability as the ability to perform a required function under specific c
onditions over a given period of time\, the most challenging aspect appear
s to be the description of the usage conditions. In particular\, the varie
ty of these conditions\, caused by country-specific road conditions and in
frastructure as well as volatile weather and traffic\, needs to be describ
ed sufficiently to recognize which requirements have to be met by the assi
st systems during their operational life.\nEspecially for the development
of heavy duty trucks\, where the execution of physical vehicle measurement
s is expensive\, electronic map data provide a powerful alternative to ana
lyse routes regarding their road characteristics\, infrastructure\, traffi
c and environmental conditions. Data generation is fast and cheap via onli
ne route planning and analysis can take place directly without using any v
ehicle resources. This presentation shows a systematic approach to classif
y heavy-duty truck routes regarding their usage conditions based on electr
onic map data and how this can be used to provide a reference stress profi
le for the reliability assessment of drive assist systems.\n\nhttps://conf
erences.enbis.org/event/11/contributions/163/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/163/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A robust method for detecting sparse changes in high-dimensional (
heteroskedastic) data
DTSTART:20210914T120000Z
DTEND:20210914T123000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-176@conferences.enbis.org
DESCRIPTION:Speakers: Inez Zwetsloot\, Zezhong Wang (City University of Ho
ng Kong)\n\nBecause of the curse-of-dimensionality\, high-dimensional proc
esses present challenges to traditional multivariate statistical process m
onitoring (SPM) techniques. In addition\, the unknown underlying distribut
ion and complicated dependency among variables such as heteroscedasticity
increase uncertainty of estimated parameters\, and decrease the effectiven
ess of control charts. In addition\, the requirement of sufficient referen
ce samples limits the application of traditional charts in high dimension
low sample size scenarios (small n\, large p). More difficulties appear wh
en detecting and diagnosing abnormal behaviors that are caused by a small
set of variables\, i.e.\, sparse changes. In this talk\, I will propose a
change-point monitoring method to detect sparse shifts in the mean vector
of high-dimensional processes. Examples from manufacturing and finance are
used to illustrate the effectiveness of the proposed method in high-dimen
sional surveillance applications.\n\nhttps://conferences.enbis.org/event/1
1/contributions/176/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/176/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Six-Sigma and Obesity – Part 2
DTSTART:20210914T104000Z
DTEND:20210914T110000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-166@conferences.enbis.org
DESCRIPTION:Speakers: Roland Caulcutt (Caulcutt Associates)\n\nWhen the Co
vid19 pandemic is no longer the prime burden on British health services\,
it might be possible to refocus on the three concerns that threatened to o
verwhelm the National Health Service in 2019. Namely\, heart disease\, ca
ncer and obesity.\nWhilst the NHS can reasonably claim to have made progre
ss with the first two\, it is faced with an ever-increasing level in obesi
ty. To non-clinical members of society this may seem rather surprising\,
considering the relative simplicity of the fat producing process\, compare
d with the extreme complexity of cancer and heart disease. It may seem ev
en more surprising to the many statisticians and process improvement profe
ssionals who witnessed the great success of blackbelts improving organisat
ional processes whilst working within a culture of “Six-Sigma”.\nPart
2 of this presentation will suggest how the blackbelt way of working can b
e adapted to improve processes within the human body. It will offer an ap
proach that might help to reduce the ever-increasing level of obesity that
has blighted so many lives.\n\nhttps://conferences.enbis.org/event/11/con
tributions/166/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/166/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A comparison of a new\, open-source graphical user interface to R
DTSTART:20210914T102000Z
DTEND:20210914T104000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-213@conferences.enbis.org
DESCRIPTION:Speakers: Matthew Metz (SKF USA Inc.)\, Bryan Dodson (SKF USA
Inc.)\, René Klerx (SKF B.V.)\n\nOrganizations\, both large and small\, h
ave a difficult time trying to standardize. In the field of statistical me
thods standardizing on a software package is especially difficult. There a
re over 50 commercial options\, over 40 open source options\, and add-ins
for spreadsheets and engineering tools. Educational licenses provide low c
osts to universities\, but graduates often find their organization does no
t use the same software they were taught at the university. One of the mos
t popular software solutions is **R**. **R** is popular because of it is f
ree\, powerful\, and covers virtually every statistical routine. Many frow
n upon **R** because it requires the user to learn scripting. There are so
me graphical user interfaces for **R**\, such as RStudio\, but these have
not met the ease-of-use level desired by most users. To address this issue
\, several leading universities have collaborated and have created a new\,
user-friendly interface for **R**. The project is called **JASP**\, and i
t is open source. This paper will demonstrate some key interfaces and capa
bilities using standard data sets for verification.\n\nhttps://conferences
.enbis.org/event/11/contributions/213/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/213/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Outlier detection in sensor networks
DTSTART:20210915T120000Z
DTEND:20210915T122000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-121@conferences.enbis.org
DESCRIPTION:Speakers: Irène GANNAZ (Univ Lyon\, INSA Lyon\, UJM\, UCBL\,
ECL\, ICJ )\, Martial AMOVIN-ASSAGBA (Arpege Master K / Université de Ly
on\, Lyon 2\, ERIC UR 3083 )\, Julien JACQUES (Université de Lyon\, Lyon
2\, ERIC UR 3083 )\n\nEmerging technologies ease the recording and collec
tion of high frequency data produced by sensor networks. From a statistica
l point of view\, these data can be view as discrete observations of rando
m functions. Our industrial goal is to detect abnormal measurement. Statis
tically\, it consists in detecting outliers in a multivariate functional d
ata set.\nWe propose a robust procedure based on contaminated mixture mode
l for both clustering and detecting outliers in multivariate functional da
ta. For each measurement\, our algorithm either classify it into one of th
e normal clusters (identifying typical normal behaviours of the sensors) o
r as an outlier.\nAn Expectation-Conditional Maximization algorithm is pr
oposed for model inference\, and its efficiency is numerically proven thro
ugh numerical experiments on simulated datasets.\nThe model is then applie
d on the industrial data set which motivated this study\, and allowed us t
o correctly detect abnormal behaviours.\n\nhttps://conferences.enbis.org/e
vent/11/contributions/121/
LOCATION:Room 5
URL:https://conferences.enbis.org/event/11/contributions/121/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Infusing Statistical Engineering at NASA
DTSTART:20210913T123000Z
DTEND:20210913T133000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-216@conferences.enbis.org
DESCRIPTION:Speakers: Peter A. Parker (NASA Langley Research Center)\n\nTh
e discipline of statistical engineering has gained recognition within NASA
by spurring innovation and efficiency\, and it has demonstrated significa
nt impact. Aerospace research and development benefits from an applicatio
n-focused statistical engineering perspective to accelerate learning\, max
imize knowledge\, ensure strategic resource investment\, and inform data-d
riven decisions. In practice\, a statistical engineering approach feature
s immersive collaboration and teaming with non-statistical disciplines to
develop solution strategies that integrate statistical methods with subjec
t-matter expertise to meet challenging research objectives. This presenta
tion provides an overview of infusing statistical engineering at NASA and
illustrates its practice through pioneering case studies in aeronautics\,
space exploration\, and atmospheric science.\n\nhttps://conferences.enbis.
org/event/11/contributions/216/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/216/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Spectral-CUSUM for Online Community Change Detection
DTSTART:20210914T140000Z
DTEND:20210914T143000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-175@conferences.enbis.org
DESCRIPTION:Speakers: Yao Xie\, Minghe Zhang\, Liyan Xie (Georgia Institut
e of Technology)\n\nDetecting abrupt structural changes in a dynamic graph
is a classic problem in statistics and machine learning. In this talk\, w
e present an online network structure change detection algorithm called sp
ectral-CUSUM to detect such changes through a subspace projection procedur
e based on the Gaussian model setting. Theoretical analysis is provided to
characterize the average run length (ARL) and expected detection delay (E
DD). Finally\, we demonstrate the good performance of the spectral-CUSUM p
rocedure using simulation and real data examples on earthquake detection i
n seismic sensor networks. This is a joint work with Minghe Zhang and Liya
n Xie.\n\nhttps://conferences.enbis.org/event/11/contributions/175/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/175/
END:VEVENT
BEGIN:VEVENT
SUMMARY:CUSUM control charts for monitoring BINARCH(1) processes
DTSTART:20210914T102000Z
DTEND:20210914T104000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-182@conferences.enbis.org
DESCRIPTION:Speakers: Maria Anastasopoulou\, Athanasios Rakitzis\n\nIn thi
s work\, we develop and study upper and lower one-sided CUSUM control char
ts for monitoring correlated counts with finite range. Often in practice\,
data of that kind can be adequately described by a first-order binomial i
nteger-valued ARCH model (or BINARCH(1)). The proposed charts are based on
the likelihood ratio and can be used for detecting upward or downward shi
fts in process mean level. The general framework for the development and t
he practical implementation of the proposed charts is given. Using Monte C
arlo simulation\, we compare the performance of the proposed CUSUM charts
with the corresponding one-sided Shewhart and EWMA charts for BINARCH(1) p
rocesses. A real-data application of the proposed charts in epidemiology i
s also discussed.\n\nhttps://conferences.enbis.org/event/11/contributions/
182/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/182/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Active coffee break: A short poll on some fun and some constructiv
e topics
DTSTART:20210914T143000Z
DTEND:20210914T144500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-232@conferences.enbis.org
DESCRIPTION:Speakers: Kristina Krebs (prognostica GmbH)\, Anja Zernig (KAI
)\n\nThis contribution offers an active coffee break. This 15-minutes acti
vity is practically an informal survey and may even be a bit funny. Togeth
er\, we create a quick picture on two selected topics within the ENBIS com
munity: a) the fashion topic Artificial Intelligence and b) ENBIS in gener
al\, which might give ideas for future events and developments within ENBI
S. Everybody is invited to take part in this short survey. Voting takes pl
ace via mentimeter - either via desktop or mobile phone - and the results
can be seen immediately. The ENBIS community’s view on these topics will
be visualised and might be published as ENBIS’ social media posts and m
ade accessible in the ENBIS Media Centre.\n\nhttps://conferences.enbis.org
/event/11/contributions/232/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/232/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Sparse abnormality detection based on variable selection for spati
ally correlated multivariate process
DTSTART:20210913T141500Z
DTEND:20210913T144500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-172@conferences.enbis.org
DESCRIPTION:Speakers: Shuai Zhang (Henan University of Engineering)\, Jian
feng Yang (Zhengzhou University)\, Jung Uk (Dongguk University)\n\nMonito
ring the manufacturing process becomes a challenging task with a huge numb
er of variables in traditional multivariate statistical process control (M
SPC) methods. However\, the rich information is often loaded with some rar
e suspicious variables\, which should be screened out and monitored. Even
though some control charts based on variable selection algorithms were pro
ven effective for dealing with such issues\, charting algorithms for the s
parse mean shift with some spatially correlated features are scarce. This
article proposes an advanced MSPC chart based on fused penalty-based varia
ble selection algorithm. First\, a fused penalised likelihood is developed
for selecting the suspicious variables. Then\, a charting statistic is em
ployed to detect potential shifts among the variables monitored. Simulatio
n experiments demonstrate that the proposed scheme can detect abnormal obs
ervation efficiently and provide root causes reasonably. It is shown that
the fused penalty can capture the spatial information and improve the robu
stness of a variables selection algorithm for spatially correlated process
.\n\nhttps://conferences.enbis.org/event/11/contributions/172/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/172/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Entropy-based Discovery of Summary Causal Graphs in Time Series
DTSTART:20210913T144500Z
DTEND:20210913T151500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-214@conferences.enbis.org
DESCRIPTION:Speakers: Eric Gaussier\, Emilie Devijver\, Ali Aït-Bachir\,
Karim ASSAAD\n\nWe address in this study the problem of learning a summary
causal graph between time series. To do so\, we first propose a new tempo
ral mutual information measure defined on a window-based representation of
time series that can detect the independence and the conditional independ
ence between two time series. We then show how this measure can be used to
derive orientation rules under the assumption that a cause cannot precede
its effect. We finally combine these two ingredients in a PC-like algorit
hm to construct the summary causal graph. This algorithm is evaluated on s
everal synthetic and real datasets that show both its efficacy and efficie
ncy.\n\nhttps://conferences.enbis.org/event/11/contributions/214/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/214/
END:VEVENT
BEGIN:VEVENT
SUMMARY:The study of variability in engineering design—An appreciation a
nd a retrospective
DTSTART:20210914T084000Z
DTEND:20210914T090000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-161@conferences.enbis.org
DESCRIPTION:Speakers: Tim Davis (We Predict Ltd. & timdavis consulting ltd
.)\n\nWe explore the concept of parameter design applied to the production
of glass beads in the manufacture of metal-encapsulated transistors. The
main motivation is to complete the analysis hinted at in the original publ
ication by Jim Morrison in 1957\, which was possibly the first example of
exploring the idea of transmitted variation in engineering design\, and a
n influential paper in the development of analytic parameter design as a s
tatistical engineering activity. Parameter design (the secondary phase of
engineering activity) is focussed on selecting the nominals of the design
variables\, to simultaneously achieve the required functional target\, wit
h minimum variance.\n\nMorrison\, SJ (1957) The study of variability in en
gineering design. Applied Statistics 6(2)\, 133–138.\n\nhttps://conferen
ces.enbis.org/event/11/contributions/161/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/161/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Attribute-Variable Alternating Inspection (AVAI): The use of $np_x
-S^2$ mixed control chart in monitoring the process variance
DTSTART:20210914T090000Z
DTEND:20210914T092000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-137@conferences.enbis.org
DESCRIPTION:Speakers: Linda Lee Ho (Universidade de São Paulo\, São Paul
o SP\, Brazil.)\, Roberto Costa Quinino (Universidade Federal de Minas Ger
ais\, Belo Horizonte MG\, Brazil)\, Leandro Alves da Silva (Universidade d
e São Paulo\, São Paulo SP\, Brazil.)\n\nThe presence of variation is an
undesirable (but natural) factor in processes. Quality improvement practi
tioners search constantly for efficient ways to monitor it\, a primary req
uirement in SPC. Generally\, inspections by attributes are cheaper and sim
pler than inspections by variables\, although they present poor performanc
e in comparison. The $S^2$ chart is widely applied in monitoring process v
ariance\, facing the need for more economical strategies that provide good
performance is the motivation of this work. Many practitioners use four t
o six units to build the $S^2$ chart\, the reduction of sample size decrea
ses their power to detect changes in process variance. This work proposes
the application of alternating inspections (by attributes and variables) u
sing sequentially samples of size $n_a$ and $n_b$ ($n_a > n_b$). The items
of sample of size $n_a$ are classified according to the $np_x$ chart proc
edure\, using a GO / NO GO gauge and counting the number of non-approved i
tems ($Y_{n_a}$). The items of sample of size $n_b$ are measured and calcu
lated its sample variance $S^2_{n_b}$. If $Y_{n_a} > UCL_{n_a}$ or $S^2_{n
_b} > UCL_{n_b}$ the process is judged out of control. The inspection alwa
ys restarts with sample size $n_a$ (using the $np_x$ chart)\, otherwise\,
the process continues. The parameters of the proposed chart are optimized
by an intensive search\, in order to outperform the $S^2$ chart (in terms
of $ARL_1$\, for a fixed $ARL_0$)\, restricted to have average sample size
closer to the sample used for $S^2$\, from their results was possible to
reduce about 10% in $ARL_1$.\n\nhttps://conferences.enbis.org/event/11/con
tributions/137/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/137/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Predicting migration patterns in Sweden using a gravity model and
neural networks
DTSTART:20210914T100000Z
DTEND:20210914T102000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-174@conferences.enbis.org
DESCRIPTION:Speakers: Magnus Pettersson\, Jonny Olofsson (Statistikkonsult
erna)\, John Pavia (Statistikkonsulterna)\n\nAccurate estimations of inter
nal migration is crucial for successful policy making and community planni
ng. This report aims to estimate internal migration between municipalities
in Sweden. \n\nTraditionally\, spatial flows of people have been modelled
using gravity models\, which assume that each region attracts or repels p
eople based on the populations of regions and distances between them. More
recently\, artificial neural networks\, which are statistical models insp
ired by biological neural networks\, have been suggested as an alternative
approach. Traditional models\, using a generalized linear framework\, hav
e been implemented and are used as a benchmark to evaluate the precision a
nd efficiency of neural network procedures.\n\nData on migration between m
unicipalities in Sweden during the years 2001 to 2020 have been extracted
from official records. There are 290 municipalities (LAU 2 according to Eu
roStat categories) in Sweden with a population size between 2 391 (Bjurhol
m) and 975 277 (Stockholm). Additional data\, including demographics and s
ocio-economics factors\, have been analyzed in an attempt to understand wh
at drives internal migration.\n\nhttps://conferences.enbis.org/event/11/co
ntributions/174/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/174/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Lessons Learned from a Career of Design of Experiments Collaborati
ons
DTSTART:20210913T151500Z
DTEND:20210913T161500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-178@conferences.enbis.org
DESCRIPTION:Speakers: Christine Anderson-Cook\n\nGeorge Box made many prof
ound and enduring theoretical and practical contributions to statistical d
esign of experiment and response surface methodology and their influence o
n industrial engineering and quality control applications. His focus on us
ing statistical tools in the right way to solving the right real-world pro
blem has been the inspiration throughout my career. Our statistical traini
ng often leads us to focus narrowly on optimality\, randomization and quan
tifying performance. However\, the practical aspects of implementation\, m
atching the design to what the experimenter really needs\, using available
knowledge about the process under study to improve the design\, and prope
r respect for the challenges of collecting data are often under-emphasized
and could undermine the success of design of experiment collaborations. I
n this talk\, I share some key lessons learned and practical advice from 1
00+ data collection collaborations with scientists and engineers across a
broad spectrum of applications.\n\nhttps://conferences.enbis.org/event/11/
contributions/178/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/178/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Influence of process parameters on part dimensional tolerances: An
Industrial Case Study
DTSTART:20210914T090000Z
DTEND:20210914T092000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-135@conferences.enbis.org
DESCRIPTION:Speakers: Narendra Akhadkar (Schneider Electric Industries)\n\
nInjection molded parts are widely used in power system protection product
s. One of the biggest challenge in an injection molding process is shrinka
ge and warpage of the molded parts. All these geometrical variations may h
ave an adverse effect on the quality of product\, functionality\, cost and
time-to-market. Our aim is to predict the spread of the functional dimens
ions and geometrical variations on the part due to variations in the input
parameters such as\, material viscosity\, packing pressure\, mold tempera
ture\, melt temperature and injection speed. \n\nThe input parameters may
vary during batch production or due to variations in the machine process s
ettings. To perform the accurate product assembly variation simulation\, t
he first step is to perform an individual part variation simulation to ren
der realistic tolerance ranges. \nWe present a method to simulate part var
iations\, coming from the input parameters variation during batch producti
on. The method is based on computer simulations and experimental validatio
n using full factorial Design of Experiments (DoE). Robustness of the simu
lation model is verified through input parameter wise sensitivity analysis
study performed using simulations and experiments\, all the results shows
a very good correlation in the material flow direction. There exists a no
n-linear interaction between material and the input process variables. It
is observed that the parameters such as\, packing pressure\, material and
mold temperature plays an import role in spread on the functional dimensio
ns and geometrical variations. This method will allow us in future to deve
lop the accurate/realistic virtual prototypes based on trusted simulated p
rocess variation.\n\nhttps://conferences.enbis.org/event/11/contributions/
135/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/135/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Priors Comparison in Bayesian mediation framework with binary outc
ome
DTSTART:20210913T141500Z
DTEND:20210913T144500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-217@conferences.enbis.org
DESCRIPTION:Speakers: Anne PHILIPPE (Nantes University)\, Jean-Michel GALH
ARRET\n\nIn human sciences\, mediation refers to a causal phenomenon in wh
ich the effect of an exposure variable 𝑋 on an outcome 𝑌 can be deco
mposed into a direct effect and an indirect effect via a third variable
𝑀 (called mediator variable). \n In mediation models\, the natural dir
ect effects and the natural indirect effects are among the parameters of i
nterest. For this model\, we construct different class of prior distributi
ons depending available information. We extend the 𝐺 -priors from the r
egression to the mediation model. We also adapt an informative transfer le
arning model to include historical information in the prior distribution.
This model will be relevant for instance in longitudinal studies with only
two or three measurement times. \nOne of the usual issues in mediation an
alysis is to test the existence of the direct and the indirect effect. Giv
en the estimation of the posterior distribution of the parameters\, we con
struct critical regions for frequentist testing process. Using simulations
\, we compare this procedure with the tests usually used in mediation anal
ysis. Finally\, we apply our approach to real data from a longitudinal stu
dy on the well-being of children in school.\n\nhttps://conferences.enbis.o
rg/event/11/contributions/217/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/217/
END:VEVENT
BEGIN:VEVENT
SUMMARY:What’s New In JMP 16
DTSTART:20210915T130000Z
DTEND:20210915T133000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-183@conferences.enbis.org
DESCRIPTION:Speakers: Chris Gotwalt\n\nJMP 16 marks a substantial expansio
n of JMP’s capabilities. In the area of DoE\, JMP 16 introduces the cand
idate set designer\, which gives the user complete control over the possib
le combinations of factor settings that will be run in the experiment. The
candidate set design capability is also very useful as an approach to Des
ign for Machine Learning\, where we use principles of optimal design to ch
oose a candidate set. JMP Pro 16 also introduces Model Screening which aut
omates fitting of a variety of machine learning models\, reducing time spe
nt in manual process of fitting and comparing various machine learning mod
els\, such as neural networks\, tree based models\, and Lasso regressions.
JMP Pro's Text Explorer platform can now perform Sentiment Analysis\, whi
ch extracts a measure of how positive or negative a document is. It also i
ntroduces Term Selection\, a patented approach to identifying words and ph
rases that are predictive of a response. The SEM platform has seen major u
pgrades in the interactivity of the path diagram. I will also introduce a
new platform called Model Screening which automates the process of finding
the best machine learning model across many different families of models\
, including neural networks\, regression trees\, the Lasso\, and much more
. Along the way\, we will also give pointers to other user useful capabili
ties that make JMP and JMP Pro 16 a powerful tool for data science and sta
tistics.\n\nhttps://conferences.enbis.org/event/11/contributions/183/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/183/
END:VEVENT
BEGIN:VEVENT
SUMMARY:The Parameter Diagram as a DoE Planning Tool
DTSTART:20210914T092000Z
DTEND:20210914T094000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-136@conferences.enbis.org
DESCRIPTION:Speakers: Matthew Barsalou\n\nStatisticians are often called u
pon to work together with Subject Matter Experts (SMEs) to perform Design
of Experiments (DoEs). The statistician may have mastered DoE\; however\,
the SME’s input may be critical in determining the correct factors\, lev
els\, and response variable of interest. The SME may be an engineer or eve
n the machine operator responsible for the daily activities at the process
that is being considered for a DoE. They may not understand what a DoE is
or what is needed for a DoE. To facilitate DoE planning\, a Parameter dia
gram (p-diagram) may be helpful. A p-diagram is not a new tool and it is o
ften used in the automotive industry for the creation of Design Failure Mo
des and Effects Analysis. The use of a p-diagram as a DoE preparation tool
\, however\, is a new application of the concept.\n\n\nThis talk will desc
ribe the p-diagram and its application in DoE. Examples will be presented
using actual DoEs from the literature. These case studies are the identifi
cation of the AA battery configuration with the longest life\, improving t
he quality of a molded part\, increasing the life of a molded tank deterre
nt device\, and the optimization of a silver powder production process. Af
ter attending this talk\, participants will be able to use a p-diagram for
DoE planning.\n\nhttps://conferences.enbis.org/event/11/contributions/136
/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/136/
END:VEVENT
BEGIN:VEVENT
SUMMARY:When\, Why and How Shewhart Control Chart Constants need to be ch
anged?
DTSTART:20210914T100000Z
DTEND:20210914T102000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-186@conferences.enbis.org
DESCRIPTION:Speakers: Vladimir Shper\, Svetlana Sheremetyeva (NUST MISiS)\
n\nShewhart Control Charts (ShCCs) are part and parcel of stability and ca
pability analysis of any process. They have long since been known and wide
ly used all over the World. The performance of ShCCs depends critically on
the values of control limits which in turn depend on the values of so-cal
led control chart constants that are considered invariable (for any given
sample size) in all SPC literature for practitioners (standards\, guides\,
handbooks\, etc.). \nOn the other hand\, many researchers proved that fo
r non-normal distribution functions (DF) the control limits may notably di
ffer from standard values. However\, there have not been even discussion a
bout changing the values of ShCCs constants yet. Meanwhile\, this is\, obv
iously\, the simplest (for practitioners) way to take the effect of non-no
rmality into consideration. \nFirstly\, we discuss what specific change of
the chart constants should be taken into account. Secondly\, we simulated
different DFs lying in different places of the well-known (β1-β2) plane
and calculated (by direct simulation) the values of the bias correction f
actors (d2\, d3\, d4) which are the basis for all chart constants. Our res
ults agree very well with the previous data\, but the further analysis sho
wed that the impact of non-normality on the ShCCs construction and interpr
etation in no way can’t be neglected. Thirdly\, we suggest rejecting the
prevalent belief of constancy of the control chart constants and explain
when and how they should be changed.\n\nhttps://conferences.enbis.org/even
t/11/contributions/186/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/186/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Autocorrelated processes in metrology with examples from ISO and J
CGM documents
DTSTART:20210913T141500Z
DTEND:20210913T144500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-184@conferences.enbis.org
DESCRIPTION:Speakers: Maurice Cox (NPL)\, Nien Fan Zhang (NIST)\n\nIt is c
ommon practice in metrology that the standard uncertainty associated with
the average of repeated observations is taken as the sample standard devia
tion of the observations divided by the square root of the sample size. Th
is uncertainty is an estimator of the standard deviation of the sample mea
n when the observations have the same mean and variance and are uncorrelat
ed.\n\nIt often happens that the observations are correlated\, especially
when data is acquired at high frequency sampling rates. In such a process\
, there are dependencies among the observations\, especially between close
ly neighbouring observations. For instance\, in continuous production such
as in the chemical industry\, many process data on quality characteristic
s are self-correlated over time. In general\, autocorrelation can be cause
d by the measuring system\, the dynamics of the process or both. \n\nFor o
bservations made of an autocorrelated process\, the uncertainty associated
with the sample mean as above is often invalid\, being inappropriately lo
w. We consider the evaluation of the standard uncertainty associated with
a sample of observations from a stationary autocorrelated process. The re
sulting standard uncertainty is consistent with relevant assumptions made
about the data generation process.\n\nThe emphasis is on a procedure that
is relatively straightforward to apply in an industrial context.\n\nExampl
es from a recent guide of the Joint Committee for Guides in Metrology and
a developing standard from the International Organization for Standardizat
ion are used to illustrate the points made.\n\nhttps://conferences.enbis.o
rg/event/11/contributions/184/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/184/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Application of the Bayesian conformity assessment framework from J
CGM 106 to lot inspection on the basis of single items
DTSTART:20210913T144500Z
DTEND:20210913T151500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-185@conferences.enbis.org
DESCRIPTION:Speakers: Steffen Uhlig (QuoData)\, Bertrand Colson (QuoData)\
n\nThe ISO 2859 and ISO 3951 series provide acceptance sampling procedures
for lot inspection\, allowing both sample size and acceptance rule to be
determined\, starting from a specific value either for the consumer or pro
ducer risk. However\, insufficient resources often prohibit the implementa
tion of “ISO sampling plans.” In cases where the sample size is alread
y known\, determined as it is by external constraints\, the focus shifts f
rom determining sample size to determining consumer and producer risks. Mo
reover\, if the sample size is very low (e.g. one single item)\, prior inf
ormation should be included in the statistical analysis. For this reason\,
it makes sense to work within a Bayesian theoretical framework\, such as
that described in JCGM 106. Accordingly\, the approach from JCGM 106 is ad
opted and broadened so as to allow application to lot inspection. The disc
ussion is based on a “real-life” example of lot inspection on the basi
s of a single item. Starting from simple assumptions\, expressions for bot
h the prior and posterior distributions are worked out\, and it is shown h
ow the concepts from JCGM 106 can be reinterpreted in the context of lot i
nspection. Finally\, specific and global consumer and producer risks are c
alculated\, and differences regarding the interpretation of these concepts
in JCGM 106 and in the ISO acceptance sampling standards are elucidated.\
n\nhttps://conferences.enbis.org/event/11/contributions/185/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/185/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Predictive Maintenance in plasma etching processes: a statistical
approach
DTSTART:20210915T124000Z
DTEND:20210915T130000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-218@conferences.enbis.org
DESCRIPTION:Speakers: Dario Casamassima (Università di Milano - Bicocca)\
, Giuseppe Fazio (StMicroelectronics)\, Riccardo Borgoni (Università di M
ilano-Bicocca)\, Diego Zappa\, Andrea Marchelli (StMicroelectronics)\, And
rea Medici (StMicroelectronics)\n\nThis contribution is a joint work of ac
ademicians and a research group of STMicroelectronics (Italy) a leading in
dustry in semiconductor manufacturing.\nThe problem under investigation re
fers to a predictive maintenance manufacturing system in Industry 4.0. Mod
ern predictive maintenance is a condition-driven preventive maintenance pr
ogram that uses possibly huge amount of data for monitoring the system to
evaluate its condition and efficiency. Machine learning and statistical le
arning techniques are nowadays the main tool by which predictive maintenan
ce operates in practice. We have tested the efficacy of such tools in the
context of plasma etching processes. More specifically the data considered
in this paper refers to an entire production cycle and had been collected
for roughly six months between December 2018 and July 2019. 2874 timepoin
ts were considered in total. Quartz degradation was monitored in terms of
the reflected power (RF). In addition to the reflected power\, the values
of more than one hundred other variables have been collected. Results sug
gest that the considered variables are related to the quartz degradation d
ifferently in different period of the production cycle. Blending different
penalized methods to shed light on the subset of covariate expected to be
prone of signals of the degradation process\, it was possible to reduce c
omplexity allowing the industrial research group to focus on them to fine
tune the best time for maintenance.\n\nhttps://conferences.enbis.org/event
/11/contributions/218/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/218/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Understanding and Addressing Complexity in Problem Solving
DTSTART:20210914T130000Z
DTEND:20210914T133000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-193@conferences.enbis.org
DESCRIPTION:Speakers: Willis Jensen (W.L. Gore & Associates)\, Jeroen De M
ast (University of Waterloo)\, Roger Hoerl (Union College)\n\nComplexity m
anifests itself in many ways when attempting to solve different problems\,
and different tools are needed to deal with the different dimensions unde
rlying that complexity. Not all complexity is created equal. We find that
most treatments of complexity in problem-solving within both the statistic
al and quality literature focus narrowly on technical complexity\, which i
ncludes the complexity of subject matter knowledge as well as complexity i
n the data access and analysis of that data. The literature lacks an under
standing of how political complexity or organizational complexity interfer
es with good technical solutions when trying to deploy a solution. Therefo
re\, people trained in statistical problem solving are ill-prepared for th
e situations they are likely to face on real projects. We propose a framew
ork that illustrates examples of complexity from our own experiences\, and
the literature. This framework highlights the need for more holistic prob
lem-solving approaches and a broader view of complexity. We also propose a
pproaches to successfully navigate complexity.\n\nhttps://conferences.enbi
s.org/event/11/contributions/193/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/193/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Deep Multistage Multi-Task Learning for Quality Prediction and Dia
gnostics of Multistage Manufacturing Systems
DTSTART:20210914T133000Z
DTEND:20210914T140000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-194@conferences.enbis.org
DESCRIPTION:Speakers: Shan Ba (LinkedIn)\, William Brenneman (Procter & Ga
mble)\, Nurretin Sergin (Arizona State University)\, Stephen Lange (the Pr
octer & Gamble)\, Hao Yan\n\nIn multistage manufacturing systems\, modelin
g multiple quality indices based on the process sensing variables is impor
tant. However\, the classic modeling technique predicts each quality varia
ble one at a time\, which fails to consider the correlation within or betw
een stages. We propose a deep multistage multi-task learning framework to
jointly predict all output sensing variables in a unified end-to-end learn
ing framework according to the sequential system architecture in the MMS.
Our numerical studies and real case study have shown that the new model ha
s a superior performance compared to many benchmark methods as well as gre
at interpretability through developed variable selection techniques.\n\nht
tps://conferences.enbis.org/event/11/contributions/194/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/194/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Tensor based Modelling of Human Motion
DTSTART:20210914T102000Z
DTEND:20210914T104000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-219@conferences.enbis.org
DESCRIPTION:Speakers: Lorena Gril\, Ulrike Kleb\, Philipp Wedenig\n\nFor
future industrial applications\, collaborative robotic systems will be a k
ey technology. A main task is to guarantee the safety of humans. To detect
hazardous situations\, commercially available robotic systems rely on dir
ect physical contact to the co-working person\, opposed to those systems e
quipped with predictive capabilities. To predict potential episodes\, wher
e the human and the robot might collide\, data of a motion tracking sensor
system are used. Based on the provided information\, the robotic system c
an avoid the unwanted physical contact by adjusting the speed or the posit
ion. A common approach of such systems is to perform human motion predicti
on by machine learning methods like Artificial Neural Networks. Our aim is
to perform human motion prediction of a repetitive assembly task by using
a Tensor-on-Tensor regression. To record human motion by means of the Opt
iTrack motion capture system\, infrared reflective markers are placed on c
orresponding joints of the human torso. The system provides unique traceab
le Cartesian coordinates (x\, y\, z) over time for each marker. Furthermor
e\, the recorded data of joint positions was transformed into the joint an
gle space to obtain the angles of joint points. To predict the human motio
n\, the contracted tensor product for the linear prediction of an outcome
array Y from the predictor array X is defined as Y = ⟨X\,B⟩ + E\, whe
re B is the coefficient tensor and E the error term. The first results are
promising for receiving multivariate predictions of highly correlated dat
a in real-time.\n\nhttps://conferences.enbis.org/event/11/contributions/21
9/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/219/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A tailored analysis of data from OMARS designs
DTSTART:20210914T094000Z
DTEND:20210914T100000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-128@conferences.enbis.org
DESCRIPTION:Speakers: Mohammed Saif Ismail Hameed (KU Leuven)\, Jose Nunez
Ares (KU Leuven)\, Peter Goos (KU Leuven (Department of Biosystems)\, Uni
versity of Antwerp (Department of Engineering Management))\n\nExperimental
data are often highly structured due to the use of experimental designs.
This does not only simplify the analysis\, but it allows for tailored met
hods of analysis that extract more information from the data than generic
methods. One group of experimental designs that are suitable for such meth
ods are the orthogonal minimally aliased response surface (OMARS) designs
(Núñez Ares and Goos\, 2020)\, where all main effects are orthogonal to
each other and to all second order effects. The design based analysis meth
od of Jones and Nachtsheim (2017) has shown significant improvement over e
xisting methods in powers to detect active effects. However\, the applicat
ion of their method is limited to only a small subgroup of OMARS designs t
hat are commonly known as definitive screening designs (DSDs). In our work
\, we not only improve upon the Jones and Nachtsheim method for DSDs\, but
we also generalize their analysis framework to the entire family of OMARS
designs. Using extensive simulations\, we show that our customized method
for analyzing data from OMARS designs is highly effective in selecting th
e true effects when compared to other modern (non-design based) analysis m
ethods\, especially in cases where the true model is complex and involves
many second order effects. \n\n\nReferences:\n\nJones\, Bradley\, and Chri
stopher J. Nachtsheim. 2017. “Effective Design-Based Model Selection for
Definitive Screening Designs.” Technometrics 59(3):319–29.\n\nNúñez
Ares\, José\, and Peter Goos. 2020. “Enumeration and Multicriteria Sel
ection of Orthogonal Minimally Aliased Response Surface Designs.” Techno
metrics 62(1):21–36.\n\nhttps://conferences.enbis.org/event/11/contribut
ions/128/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/128/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Causal Rules Extraction in Time Series Data
DTSTART:20210913T134500Z
DTEND:20210913T141500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-196@conferences.enbis.org
DESCRIPTION:Speakers: Amin Dhaou\, Sebastien Gourvenec (TotalEnergies)\, E
rwan Le Pennec (CMAP\, Ecole Polytechnique\, Institut Polytechnique de Par
is\, France)\, Antoine Bertoncello (TotalEnergies)\, Josselin Garnier (CMA
P - Ecole Polytechnique)\n\nThe number of complex infrastructures in an in
dustrial setting\nis growing and is not immune to unexplained recurring ev
ents\nsuch as breakdowns or failure that can have an economic and\nenviron
mental impact. To understand these phenomena\, sensors\nhave been placed o
n the different infrastructures to track\, monitor\,\nand control the dyna
mics of the systems. The causal study of these\ndata allows predictive and
prescriptive maintenance to be carried\nout. It helps to understand the a
ppearance of a problem and find\ncounterfactual outcomes to better operate
and defuse the event.\nIn this paper\, we introduce a novel approach comb
ining the\ncase-crossover design which is used to investigate acute trigge
rs\nof diseases in epidemiology\, and the Apriori algorithm which is a\nda
ta mining technique allowing to find relevant rules in a dataset.\nThe res
ulting time series causal algorithm extracts interesting rules\nin our app
lication case which is a non-linear time series dataset.\nIn addition\, a
predictive rule-based algorithm demonstrates the\npotential of the propose
d method.\n\nhttps://conferences.enbis.org/event/11/contributions/196/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/196/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A Digital Twin Approach for Statistical Process Monitoring of a Hi
gh-Dimensional Microelectronic Assembly Process
DTSTART:20210915T122000Z
DTEND:20210915T124000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-191@conferences.enbis.org
DESCRIPTION:Speakers: Marco P. Seabra dos Reis (University of Coimbra\, De
partment of Chemical Engineering)\, Tiago Rato\, Ricardo Rendall (Dow Chem
ical Co.)\, Pedro Delgado (Bosch Car Multimedia\, SA)\, Cristina Martins (
Bosch Car Multimedia\, SA)\n\nWe address a real case study of Statistical
Process Monitoring (SPM) of a Surface Mount Technology (SMT) production li
ne at Bosch Car Multimedia\, where more than 17 thousand product variables
are collected for each product. The basic assumption of SPM is that all r
elevant “common causes” of variation are represented in the reference
dataset (Phase 1 analysis). However\, we argue and demonstrate that this a
ssumption is often not met\, namely in the industrial process under analys
is. Therefore\, we derived a digital twin from first principles modeling o
f the dominant modes of common cause variation. With such digital twin\, i
t is possible to enrich the historical dataset with simulated data represe
nting a comprehensive coverage of the actual operational space. This metho
dology avoids the excessive false alarm problem that affected the unit and
that prevented the use of SPM. We also show how to compute the monitoring
statistics and set their control limits\, as well as to conduct fault dia
gnosis when an abnormal event is detected.\n\nhttps://conferences.enbis.or
g/event/11/contributions/191/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/191/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Prediction intervals for real estate price prediction
DTSTART:20210914T084000Z
DTEND:20210914T090000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-151@conferences.enbis.org
DESCRIPTION:Speakers: Rainer Göb\, Moritz Beck (University of Wuerzburg)\
n\nAutomated procedures of real estate price estimation and prediction hav
e been used in the real estate sector since 15 years. Various providers of
real estate price predictions are available\, e. g.\, the platform Zillow
\, or Immoscout 24 from Germany. Simultaneously\, the problem of real esta
te price prediction has become a subject of statistical and machine learni
ng literature. The current providers and theory strongly focus on point pr
edictions. For users\, however\, interval predictions are more useful and
reliable. A perspective approach for obtaining prediction intervals is qua
ntile regression. We analyse several methods of quantile regression\, in p
articular linear quantile regression\, support vector quantile regression\
, quantile gradient boosting\, quantile random forest\, $k$-nearest neighb
our quantile regression\, $L_1$-norm quantile regression. The performance
of the methods are evaluated on a large data set of real estate prices wit
h relevant covariates. It turns out that the best predictive power is obta
ined by linear quantile regression and $k$-nearest neighbour quantile regr
ession.\n\nhttps://conferences.enbis.org/event/11/contributions/151/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/151/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Strategies for Supersaturated Screening: Group Orthogonal and Cons
trained Var(s) Designs
DTSTART:20210914T140000Z
DTEND:20210914T143000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-195@conferences.enbis.org
DESCRIPTION:Speakers: Byran Smucker (Miami University)\, David Edwards (Vi
rginia Commonwealth University)\, Jonathan Stallrich (NC State University)
\, Maria Weese (Miami University)\n\nDespite the vast amount of literature
on supersaturated designs (SSDs)\, there is a scant record of their use i
n practice. We contend this imbalance is due to conflicting recommendatio
ns regarding SSD use in the literature as well as the designs' inabilities
to meet practitioners' analysis expectations. To address these issues\, w
e first summarize practitioner concerns and expectations of SSDs as determ
ined via an informal questionnaire. Next\, we discuss and compare two rece
nt SSDs that pair a design construction method with a particular analysis
method. The choice of a design/analysis pairing is shown to depend on the
screening objective. Group orthogonal supersaturated designs\, when paire
d with our new\, modified analysis\, are demonstrated to have high power e
ven with many active factors. Constrained positive Var(s)-optimal designs\
, when paired with the Dantzig selector\, are recommended when effect dire
ctions can be reasonably specified in advance\; this strategy reasonably c
ontrols type 1 error rates while still identifying a high proportion of ac
tive factors.\n\nhttps://conferences.enbis.org/event/11/contributions/195/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/195/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Modern Methods of Quantifying Parameter Uncertainties via Bayesian
Inference
DTSTART:20210915T140000Z
DTEND:20210915T143000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-201@conferences.enbis.org
DESCRIPTION:Speakers: Nando Farchmin (Physikalisch-Technische Bundesanstal
t)\, Sebastian Heidenreich (Physikalisch-Technische Bundesanstalt)\, Maren
Casfor Zapata (Physikalisch-Technische Bundesanstalt)\n\nIn modern metrol
ogy an exact specification of unknown characteristic values\, such as shap
e parameters or material constants\, is often not possible due to e.g. the
ever decreasing size of the objects under investigation. Using non-destru
ctive measurements and inverse problems is both an elegant and economical
way to obtain the desired information while also providing the possibility
to determine uncertainties of the reconstructed parameter values. In this
talk we present state-of-the-art approaches to quantify these parameter u
ncertainties by Bayesian inference. Among others\, we discuss surrogate ap
proximations for high-dimensional problems to circumvent computationally d
emanding physical models\, error correction via the introduction of an add
itional model error to automatically correct systematic model discrepancie
s and transport of measure approaches using invertible neural networks whi
ch accelerate sampling from the problem posterior drastically in compariso
n to standard MCMC strategies. The presented methods are illustrated by ap
plications in optical shape reconstruction of nano-structures\, in particu
lar photo-lithography masks\, with scattering and grazing incidence X-ray
fluorescence measurements.\n\nhttps://conferences.enbis.org/event/11/contr
ibutions/201/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/201/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Cleanliness an underestimated area when solving problems on Safety
Critical Aerospace parts
DTSTART:20210914T100000Z
DTEND:20210914T102000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-187@conferences.enbis.org
DESCRIPTION:Speakers: Sören Knuts\n\nCleaning is a method that has standa
rds and specifications within Aerospace industry of how to fulfil a cleani
ng requirement with respect to a certain material. Nevertheless\, it is an
area where underlying technical problems tend to be of an intermittent an
d long-term nature. Cause and effect-wise relationships are hard to derive
that makes the problem solving more of a guessing game. The lack of under
standing of the underlying mechanisms of how the cleaning method is intera
cting with the material\, is limiting the C&E-analysis and makes it almost
impossible to reach common understanding of how-to priorities improvement
initiatives in the cross functional product team. This is even further ha
mpered by the lack of a precise measurement system and standardized proced
ures of how to evaluate the capability of the measurements relative cleani
ng variations on a regular basis. A measurement system including visualiza
tion methods that not only detects bad performances of the cleaning method
but is also monitors its nominal performance within limits over time\, th
at is\, control limits. \nIn this presentation a technical cleanliness p
roblem related to background fluorescence on a safety critical aero engine
part is shown. The background fluorescence limits the inspectability of t
he part\, and further cleaning must be done on the part in order to make i
t possible to inspect the part. The fuzzy origin and different hypothesis
are discussed\, and the way to attack the difficulty of measurement proble
m is also discussed.\n\nhttps://conferences.enbis.org/event/11/contributio
ns/187/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/187/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Bayesian I-optimal designs for choice experiments with mixtures
DTSTART:20210914T084000Z
DTEND:20210914T090000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-124@conferences.enbis.org
DESCRIPTION:Speakers: Peter Goos (KU Leuven)\, Mario Becerra (KU Leuven)\n
\nDiscrete choice experiments are frequently used to quantify consumer pre
ferences by having respondents choose between different alternatives. Choi
ce experiments involving mixtures of ingredients have been largely overloo
ked in the literature\, even though many products and services can be desc
ribed as mixtures of ingredients. As a consequence\, little research has b
een done on the optimal design of choice experiments involving mixtures. T
he only existing research has focused on D-optimal designs\, which means t
hat an estimation-based approach was adopted. However\, in experiments wit
h mixtures\, it is crucial to obtain models that yield precise predictions
for any combination of ingredient proportions. This is because the goal o
f mixture experiments generally is to find the mixture that optimizes the
respondents' utility. As a result\, the I-optimality criterion is more sui
table for designing choice experiments with mixtures than the D-optimality
criterion because the I-optimality criterion focuses on getting precise p
redictions with the estimated statistical model. In this paper\, we study
Bayesian I-optimal designs\, compare them with their Bayesian D-optimal co
unterparts\, and show that the former designs perform substantially better
than the latter in terms of the variance of the predicted utility.\n\nhtt
ps://conferences.enbis.org/event/11/contributions/124/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/124/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Bayesian Transfer Learning for modelling the hydrocracking process
using kriging
DTSTART:20210914T084000Z
DTEND:20210914T090000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-146@conferences.enbis.org
DESCRIPTION:Speakers: Julien JACQUES (Université de Lyon\, Lyon 2\, ERIC
UR 3083 )\, Victor LAMEIRAS FRANCO DA COSTA\, Benoit CELSE\, Loïc IAPTEF
F\n\nHydrocracking process reaction takes place in presence of a catalyst\
, and when supplying a catalyst\, a vendor must guarantee its performance.
In this work\, the linear and the kriging model are considered to model t
he process. The construction of predictive models is based on experimental
data and experiments are very expensive. New catalysts are constantly bei
ng developed so that each new generation of a catalyst requires a new mode
l that is until now built from scratch from new experiments. The aim of th
is work is to build the best predictive model for a new catalyst from fewe
r observations and using the observations of previous generation catalysts
. This task is known as transfer learning.\n\nThe method used is the trans
fer knowledge of parameters approach\, which consists in transferring regr
ession models from an old dataset to a new one. \nIn order to adapt the pa
st knowledge to the new catalyst\, a Bayesian approach is considered. The
idea of the approach is to take as prior a distribution centered on the pr
evious model parameters. A pragmatic approach to chose the prior variance
ensuring that it is large enough to allow parameter change and small enoug
h to retain the information is proposed.\n\nWith the Bayesian transfer app
roach\, the RMSE scores for the transferred models are always lower than t
hose obtained without transfer\, especially when the number of observation
s is low. Satisfactory models can be fitted with only five new observation
s. Without transfer\, reaching the same model quality requires about fifty
observations.\n\nhttps://conferences.enbis.org/event/11/contributions/146
/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/146/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Machine Learning Approach to Predict Land Prices using Spatial Dep
endency Factors
DTSTART:20210915T122000Z
DTEND:20210915T124000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-199@conferences.enbis.org
DESCRIPTION:Speakers: Supun Delpagoda (Department of Physical Sciences\, F
aculty of Applied Sciences\, Rajarata University of Sri Lanka.)\, Kaushaly
a Premachandra (Department of Physical Sciences\, Faculty of Applied Scien
ces\, Rajarata University of Sri Lanka.)\, Ranjan Dissanayake (Department
of Physical Sciences\, Faculty of Applied Sciences\, Rajarata University o
f Sri Lanka.)\n\nIn real estate models\, spatial variation is an important
factor in predicting land prices. Spatial dependency factors (SDFs) under
spatial variation play a key role in predicting land prices. The objectiv
e of this study was to develop a novel real estate model that is suitable
for Sri Lanka by exploring the factors affecting the prediction of land pr
ices using ordinary least squares regression (OLS) and artificial neural n
etworks (ANNs). For this purpose\, a total of 1000 samples on land prices
(dependent variable) were collected from the Kesbewa Division in Colombo m
etropolitan city\, using various web commercials\, and explored spatial de
pendency factors (independent variable) such as distance from the particu
lar land to the nearest main road\, city\, public or private hospital and
school. The real estate model was developed and validated using the SDFs
that were calculated using Google Maps and R-Studio. The OLS model showed
that SDFs have a significant effect on land pricing $(p\n\nhttps://confere
nces.enbis.org/event/11/contributions/199/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/199/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Estimating the Time to Reach the Curing Temperature in Autoclave C
uring Processes
DTSTART:20210914T090000Z
DTEND:20210914T092000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-154@conferences.enbis.org
DESCRIPTION:Speakers: Gözdenur Kırdar (Hacettepe University)\, Murat Can
er Testik (Hacettepe University)\, Diclehan Tezcaner Öztürk\n\nAutoclave
curing process is one of the important stages in manufacturing. In this p
rocess\, multiple parts are loaded in the autoclave as a batch\, they are
heated up to their curing temperature (heating phase) and cured at that te
mperature for their dwell period. There are two main considerations that a
ffect how parts are placed in the autoclave. Firstly\, if some parts reach
the curing temperature earlier than the others\, they are overcured until
the remaining parts reach that temperature. This overcuring worsens the q
uality of the final products. Secondly\, shorter curing cycles are preferr
ed to increase productivity of the whole system. Both considerations can b
e addressed if the time required for each part to reach the curing tempera
ture (heating time) is known in advance. However\, there are no establishe
d relationships between part properties and their heating times. In this s
tudy\, we develop the relation between part and batch properties with the
heating times. We consider the effects of location\, part weight\, part si
ze\, and batch properties on the heating times. The autoclave charge floor
is imaginarily divided in 18 areas and for each area multiple linear regr
ession models that estimate the heating times are developed. Additionally\
, a biobjective optimization model is developed that finds efficient place
ments of parts\, minimizing the maximum overcuring duration and the durati
on of the heating phase. The approach is applied on a real case\, and an e
fficient solution is implemented. The regression models result in signific
antly close estimations to the realizations.\n\nhttps://conferences.enbis.
org/event/11/contributions/154/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/154/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Modelling electric vehicle charging load with point processes and
multivariate mixtures
DTSTART:20210914T092000Z
DTEND:20210914T094000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-181@conferences.enbis.org
DESCRIPTION:Speakers: Yannig Goude (EDF R&D)\, Hui Yan (EDF R&D)\, Jean-Mi
chel Poggi (Université Paris-Saclay)\, Pascal Massart (Universite Paris-S
aclay)\, Yvenn Amara-Ouali\n\nNumerous countries are making electric vehic
les their key priority to reduce emissions in their transport sector. This
emerging market is subject to multiple unknowns and in particular the cha
rging behaviours of electric vehicles. The lack of data describing the int
eractions between electric vehicles and charging points hinders the develo
pment of statistical models describing this interaction [1]. In this work\
, we want to address this gap by proposing a data-driven model of the elec
tric vehicle charging load benchmarked on open charging session datasets.
These open datasets cover all common charging behaviours: (a) public charg
ing\, (b) workplace charging\, (c) residential charging. The model introdu
ced in this work focuses on three variables that are paramount for reconst
ructing the electric vehicle charging load in an uncontrolled charging env
ironment: the arrival time\, the charging duration\, and the energy demand
ed for each charging session. The arrivals of EVs at charging points are c
haracterised by as a non-homogenous Poisson Process\, and the charging dur
ation and energy demanded are modelled conditionally to these arrival time
s as a bivariate mixture of Gaussian distributions. We compare the perform
ances of the model proposed on all these datasets across different metrics
.\n[1] Amara-Ouali\, Y. et al. 2021. A Review of Electric Vehicle Load Ope
n Data and Models. Energies. 14\, 8 (Apr. 2021)\, 2233. DOI:https://doi.or
g/10.3390/en14082233.\n\nhttps://conferences.enbis.org/event/11/contributi
ons/181/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/181/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Heteroscedastic Gaussian Process regression for assessing interpol
ation uncertainty of essential climate variables
DTSTART:20210913T144500Z
DTEND:20210913T151500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-202@conferences.enbis.org
DESCRIPTION:Speakers: Pietro Colombo\, Alessandro Fasso' (University of Be
rgamo)\n\nRecent advancements\, [2][3]\, in interpolation uncertainty esti
mation for the vertical profiles of ECV (essential climate variables)\, ha
ve shown the Gaussian process regression to be a valid interpolator. Gauss
ian process regression assumes the variance to be constant along the atmos
pheric profile. This behaviour is known as the homoscedasticity of the res
iduals.\nHowever\, climate variables often present heteroscedastic residua
ls. The implementation of Gaussian process regression that accounts for th
is latter aspect is a plausible way to improve the interpolation uncertain
ty estimation. In [2]\, these authors recently showed that Gaussian Proces
s regression gives an effective interpolator for relative humidity measure
ments\, especially when the variability of underlining natural process is
high. \nIn this talk\, we consider Gaussian methods that allow for heteros
cedasticity\, e.g. [1]\, hence handling situations in which we have input-
dependent variance. In this way\, we will provide a more precise estimate
of the interpolation uncertainty.\n\nReferences\n\n\n\n[1] Wang C.\, (2014
) Gaussian Process Regression with Heteroscedastic Residuals and Fast MCMC
Methods\, PhD thesis\, Graduate Department of Statistics\, University of
Toronto.\n[2] Colombo\, P.\, and Fassò A.\, (2021) Joint Virtual Workshop
of ENBIS and MATHMET Mathematical and Statistical Methods for Metrology\,
MSMM 2021.\n[3] Fassò\, A.\, Michael S.\, and von Rohden C. (2020) "Inte
rpolation uncertainty of atmospheric temperature profiles."\, Atmospheric
Measurement Techniques\, 13(12): 6445-6458.\n\nhttps://conferences.enbis.o
rg/event/11/contributions/202/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/202/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Randomizing versus not randomizing split-plot experiments
DTSTART:20210913T134500Z
DTEND:20210913T141500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-167@conferences.enbis.org
DESCRIPTION:Speakers: G. Geoffrey Vining (Department of Statistics\, Virg
inia Tech)\, Francesco Bertocci (Department of Global Transducer Technolog
y\, Esaote S.p.A)\, Nedka Dechkova Nikiforova (Department of Statistics Co
mputer Science Applications "G. Parenti"\, University of Florence)\, Rosse
lla Berni\n\nRandomization is a fundamental principle underlying the stati
stical planning of experiments. In this talk\, we illustrate the impact wh
en the experimenter either cannot or chooses not to randomize the applicat
ion of the experimental factors to their appropriate experimental units fo
r split-plot experiments (Berni et al.\, 2020). The specific context is an
experiment to improve the production process of an ultrasound transducer
for medical imaging. Due to the constraints presented by the company requi
rements\, some of the design factors cannot be randomized. Through a simul
ation study based on the experiment for the transducer\, we illustrate vis
ually the impact of a linear trend over time for both the randomized and n
onrandomized situations\, at the whole-plot and at the sub-plot levels. We
assess the effect of randomizing versus not randomizing by considering th
e estimated model coefficients\, and the whole-plot and sub-plot residuals
. We also illustrate how to detect and to estimate the linear trend if the
design is properly randomized\, by also analyzing the impact of different
slopes for the trend. We show that the nonrandomized design cannot detect
the presence of the linear trend through residual plots because the impac
t of the trend is to bias the estimated coefficients. The simulation study
provides an excellent way to explain to engineers and practitioners the f
undamental role of randomization in the design and analysis of experiments
. \nREFERENCES:\nRossella Berni\, Francesco Bertocci\, Nedka D. Nikiforova
& G. Geoffrey Vining (2020) A tutorial on randomizing versus not randomiz
ing Split-Plot experiments\, Quality Engineering\, 32:1\, 25-45\, DOI: 10.
1080/08982112.2019.1617422.\n\nhttps://conferences.enbis.org/event/11/cont
ributions/167/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/167/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A hybrid method for degradation assessment and fault detection in
rolling element bearings
DTSTART:20210914T094000Z
DTEND:20210914T100000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-190@conferences.enbis.org
DESCRIPTION:Speakers: Jacob Bortman\, Rosenblatt Jonathan\, Renata Klein\,
Yonatan Nissim\n\nRolling Element Bearings (REBs) are key components in r
otary machines\, e.g.\, turbines and engines. REBs tend to suffer from var
ious faults causing serious damage to the whole system. Therefore\, many t
echniques and algorithms have been developed over the past years\, to dete
ct and diagnose\, as early as possible\, an incipient fault and its propag
ation using vibration monitoring. Moreover\, some of the methods attempt t
o estimate the severity of the degrading system\, to achieve better progno
stics and Remaining Useful Life (RUL) estimation.\nWhile data-driven metho
ds\, such as machine and deep learning continue to grow\, they still lack
physical awareness and are yet sensitive to some phenomena not related to
the fault. In this paper\, we present a hybrid method for REBs fault diagn
osis which includes physics-based pre-processing techniques combined with
deep learning models for a semi-supervised fault detection. To compare and
evaluate our results\, we also compare performance of different detection
methods on data from an endurance test with a propagating fault in the ou
ter race. The methods we compare are both from physics-based and data-driv
en fields. The results show that the presented hybrid method including phy
sical-aware signal processing techniques and feature extraction related to
the bearing fault\, can increase the reliability and interpretability of
the data-driven model. The health indicator received from the proposed met
hod showed better trendiness indicating the severity of the fault and impr
oved the health track of the degrading system.\n\nhttps://conferences.enbi
s.org/event/11/contributions/190/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/190/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Addressing statistics and data science educational challenges wit
h simulation platforms
DTSTART:20210915T120000Z
DTEND:20210915T122000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-119@conferences.enbis.org
DESCRIPTION:Speakers: Ron Kenett (KPA Group and Samuel Neaman Institute\,
Technion\, Israel)\, Chris Gotwalt (JMP Division\, SAS\, Research Triangle
)\n\nComputer age statistics\, machine learning and\, in general\, data an
alytics is having an ubiquitous impact on industry\, business and services
. This data transformation requires a growing workforce which is up to the
job in terms of knowledge\, skills and capabilities. The deployment of an
alytics needs to address organizational needs\, invoke proper methods\, bu
ild on adequate infrastructures and providing the right skills to the righ
t people. The talk will show how embedding simulations in analytic platfor
ms can provide an efficient educational experience to both students\, in c
olleges and universities\, and company employees engaged in lifelong learn
ing initiatives. Specifically\, we will show how a simulator\, such as the
ones provided in https://intelitek.com/\, can be used to learn tools inv
oked in monitoring\, diagnostic\, prognostic and prescriptive analytics. W
e will also emphasize that such upskilling requires a focus on conceptual
understanding affecting both the pedagogical approach and the learning ass
essment tools. The topics covered\, from an educational perspective includ
e information quality\, data science\, industrial statistics\, hybrid teac
hing\, simulations and conceptual understanding. Throughout the presentati
on\, the JMP platform (www.jmp.com ) will be used to demonstrate the point
s made in the talk.\n\nReference\n• Marco Reis & Ron S. Kenett (2017) A
structured overview on the use of computational simulators for teaching st
atistical methods\, Quality Engineering\, 29:4\, 730-744.\n\nhttps://confe
rences.enbis.org/event/11/contributions/119/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/119/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Enumeration of large mixed four-and-two-level regular designs
DTSTART:20210915T100000Z
DTEND:20210915T102000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-126@conferences.enbis.org
DESCRIPTION:Speakers: Alexandre Bohyn (KU Leuven)\, Peter Goos (KU Leuven)
\, Eric Schoen (KU Leuven)\n\nA protocol for a bio-assay involves a substa
ntial number of steps that may affect the end result. To identify the infl
uential steps\, screening experiments can be employed with each step corre
sponding to a factor and different versions of the step corresponding to f
actor levels. The designs for such experiments usually include factors wit
h two levels only. Adding a few four-level factors would allow inclusion o
f multi-level categorical factors or quantitative factors that may show qu
adratic or even higher-order effects. However\, while a reliable investiga
tion of the vast number of different factors requires designs with larger
run sizes\, catalogs of designs with both two-level factors and four-level
factors are only available for up to 32 runs. In this presentation\, we d
iscuss the generation of such designs. We use the principles of **extensio
n** (adding columns to an existing design to form candidate designs) and *
*reduction** (removing equivalent designs from the set of candidates). Mor
e specifically\, we select three algorithms from the current literature fo
r the generation of complete sets of two-level designs\, adapt them to enu
merate designs with both two-level and four-level factors\, and compare th
e efficiency of the adapted algorithms for generating complete sets of non
-equivalent designs. Finally\, we use the most efficient method to generat
e a complete catalog of designs with both two-level and four-level factors
for run sizes 32\, 64\, 128 and 256.\n\nhttps://conferences.enbis.org/eve
nt/11/contributions/126/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/126/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A novel fault detection and diagnosis approach based on orthogonal
autoencoders
DTSTART:20210915T100000Z
DTEND:20210915T102000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-132@conferences.enbis.org
DESCRIPTION:Speakers: Davide Cacciarelli (Technical University of Denmark
(DTU))\, Murat Kulahci (Technical University of Denmark (DTU))\n\nThe need
to analyze complex nonlinear data coming from industrial production setti
ngs is fostering the use of deep learning algorithms in Statistical Proces
s Control (SPC) schemes. In this work\, a new SPC framework based on ortho
gonal autoencoders (OAEs) is proposed. A regularized loss function ensures
the invertibility of the covariance matrix when computing the Hotelling $
T^2$ statistic and non-parametric upper control limits are obtained from a
kernel density estimation. When an out-of-control situation is detected\,
we propose an adaptation of the integrated gradients method to perform a
fault contribution analysis by interpreting the bottleneck of the network.
The performance of the proposed method is compared with traditional appro
aches like principal component analysis (PCA) and Kernel PCA (KPCA). In th
e analysis\, we examine how the detection performances are affected by cha
nges in the dimensionality of the latent space. Determining the right dime
nsionality is a challenging problem in SPC since the models are usually tr
ained on phase I data solely\, with little to no prior knowledge on the tr
ue latent structure of the underlying process. Moreover\, data containing
faults is quite scarce in industrial settings\, reducing the possibility t
o perform a thorough investigation on the detection performances for diffe
rent numbers of extracted features. The results show how OAEs offer robust
results despite radical changes in the latent dimension while the detecti
on performances of traditional methods witness significant fluctuations.\n
\nhttps://conferences.enbis.org/event/11/contributions/132/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/132/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Dubious new control chart designs — a disturbing trend
DTSTART:20210914T100000Z
DTEND:20210914T102000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-144@conferences.enbis.org
DESCRIPTION:Speakers: William Woodall (Virginia Tech\, Blacksburg VA\, USA
)\, Sven Knoth (Helmut Schmidt University Hamburg\, Germany)\, Marzieh Kha
kifirooz (Tecnologico de Monterrey\, Monterrey\, Nuevo Leon\, Mexico)\, Vi
ctor Tercero-Gomez (Tecnologico de Monterrey\, Monterrey\, Nuevo Leon\, Me
xico)\n\nFor the last twenty years\, a plethora of new ``memory-type'' con
trol charts have been proposed. They share some common features: (i) decep
tively good zero-state average run-length (ARL) performance\, but poor ste
ady-state performance\, (ii) design\, deployment and analysis significantl
y more complicated than for established charts\, (iii) comparisons made to
unnecessarily weak competitors\, and (iv) resulting weighting of the obse
rved data overemphasizing the distant past. For the most prominent represe
ntative\, the synthetic chart\, these problems have been already discussed
(Davis/Woodall 2002\; Knoth 2016)\, but these and other approaches contin
ue to gain more and more popularity despite their substantial weaknesses.
Recently\, Knoth et al. (2021a\,b) elaborated on issues related to the PM\
, HWMA\, and GWMA charts. Here\, we want to give an overview on this contr
ol chart jumble. We augment the typical zero-state ARL analysis by calcula
ting the more meaningful conditional expected delay (CED) values and their
limit\, the conditional steady-state ARL. Moreover\, we select the compet
itor (EWMA) in a more reasonable way. It is demonstrated that in all cases
the classical chart should be preferred. The various abbreviations (DEWMA
... TEWMA) will be explained during the talk. \n\nDAVIS\, WOODALL (2002).
\n”Evaluating and Improving the Synthetic Control Chart”.\nJQT 34(2)\,
200–208.\n\nKNOTH (2016).\n”The Case Against the Use of Synthetic Con
trol Charts”.\nJQT\, 48(2)\, 178–195.\n\nKNOTH\, TERCERO-GÓMEZ\, KHAK
IFIROOZ\, WOODALL (2021a).\n”The Impracticality of Homogeneously Weighte
d Moving Average and Progressive Mean\nControl Chart Approaches”.\nTo ap
pear in QREI.\n\nKNOTH\, WOODALL\, TERCERO-GÓMEZ (2021b).\n”The Case ag
ainst Generally Weighted Moving Average (GWMA) Control Charts”. Submitte
d.\n\nhttps://conferences.enbis.org/event/11/contributions/144/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/144/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Fault detection in continuous chemical processes using a PCA-based
local approach
DTSTART:20210915T102000Z
DTEND:20210915T104000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-160@conferences.enbis.org
DESCRIPTION:Speakers: Gustavo Almeida (Federal University of Minas Gerais)
\, Danilo Reis (Federal University of Minas Gerais)\, Gabriela Costa (Fede
ral University of Minas Gerais)\, Gustavo Barcelos (Federal University of
Minas Gerais)\, Leticia Almada (Federal University of Minas Gerais)\n\nEar
ly fault detection in the process industry is crucial to mitigate potentia
l impacts. Despite being widely studied\, fault detection remains a practi
cal challenge. Principal components analysis (PCA) has been commonly used
for this purpose. This work employs a PCA-based local approach to improve
fault detection efficiency. This is done by adopting individual control li
mits for the principal components. Several numbers of retained components
(d = [5:45]\, in steps of 5) were investigated. The false alarm rate (FAR)
was set at 1%. The level of significance () for the control limits was
a function of d. The well-known Tennessee benchmark was used as the case
study\, whose faults can be grouped into easy\, intermediate\, hard and ve
ry hard detection faults. Significant improvements were reached for the in
termediate and hard groups in comparison to the classic use of PCA. Relati
ve gains around 50% in MDR (missed detection rate) were obtained for two o
ut of the three intermediate faults\, given the T2 statistic. In the hard
to detect group\, all six faults except one presented relative gain in MDR
above 50% for both statistics T2 and Q. In general\, the local approach w
as superior for 16\, equivalent for 2\, and inferior for 3 (easy detection
faults) faults given T2. These values were\, respectively\, equal to 11\,
5 and 5 (four easy and one intermediate detection faults)\, for the Q sta
tistic. The overall results suggest that the local approach was more prone
to detect more difficult faults\, which is of most interest in practice.\
n\nhttps://conferences.enbis.org/event/11/contributions/160/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/160/
END:VEVENT
BEGIN:VEVENT
SUMMARY:PREDICTION OF PRECIPITATION THROUGH WEATHER VARIABLES BY FUNCTIONA
L REGRESSION MODELS
DTSTART:20210915T100000Z
DTEND:20210915T102000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-134@conferences.enbis.org
DESCRIPTION:Speakers: Javier Tarrío Saavedra (Grupo MODES\, CITIC\, ITMAT
I\, Department of Mathematics\, Escola Politécnica Superior\, Universidad
e da Coruña)\, Salvador Naya (Grupo MODES\, CITIC\, ITMATI\, Department o
f Mathematics\, Escola Politécnica Superior\, Universidade da Coruña)\,
Danilo Leandro Loza Quispillo (Departamento de Matemática\, Escuela Pol
itécnica Nacional)\, Angel Omar Llambo Delgado (Departamento de Matemát
ica\, Escuela Politécnica Nacional)\, Miguel Alfonso Flores Sánchez (G
rupo MODES\, SIGTI\, FADE\, Departamento de Matemática\, Escuela Politéc
nica Nacional)\n\nIn this work\, we are going to predict precipitation thr
ough the use of different functional regression models (FRM) and the best
fit is selected between: Functional Linear Model with Basic Representation
(FLR)\, Functional Linear Model with Functional Basis by Principal Compon
ents (PC)\, Functional Linear Model with Functional Basis of Principal Com
ponents by Partial Least Squares (PLS) and the adaptation of a Functional
Linear Model with two independent variables.\n\nThe results obtained by th
ese models are very useful to understand the behavior of precipitation. Wh
en compare the results it is deduced that the functional regression model
that includes two explanatory functional variables provides a better fit s
ince the variation of precipitation is explained to through temperature an
d wind speed by 91%. Finally\, with this model\, tests are carried out tha
t allow the stability of its parameters to be analyzed.\n\nThis study allo
ws us to establish meteorological parameters that help us to illustrate sc
enarios (favorable and adverse) in order to better cope with the temporal
that arise during the year\, so that projects or studies can be put into p
ractice that allow improving socioeconomic conditions of the agricultural
sector.\n\nhttps://conferences.enbis.org/event/11/contributions/134/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/134/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Non-parametric multivariate control charts based on data depth not
ion
DTSTART:20210914T133000Z
DTEND:20210914T140000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-169@conferences.enbis.org
DESCRIPTION:Speakers: Giuseppe Pandolfo (University of Naples Federico II)
\, Carmela Iorio (University of Naples Federico II)\n\nA control chart is
used to monitor a process variable over time by providing information abou
t the process behavior. Monitoring the process of related variables is usu
ally called a multivariate quality control problem. Multivariate control c
harts\, needed when dealing with more than one quality variable\, relies o
n very specific models for the data generating process. When large histori
cal data set are available\, previous knowledge of the process may not be
available or a unique model for all the features cannot be adopted\, and n
o specific parametric model turns out to be appropriate and some alternati
ve solutions should be adopted. Hence\, exploiting non-parametric methods
to build a control chart appears a reasonable choice. Non-parametric contr
ol charts require no distributional assumptions on the process data and ge
nerally enjoy more robustness\, i.e. are less sensitive to outlier\, over
parametric control schemes. Among the possible non-parametric statistical
techniques\, data depth functions are gaining a growing interest in multiv
ariate quality control. These are nonparametric functions which are able t
o provide a dimension reduction to high-dimensional problems. Several dept
h measures are effective for purposes\, even in the case of deviation from
the normality assumption. However\, the use of the L^p data depth for con
structing nonparametric multivariate control charts has been neglected so
far. Hence\, the contribution of this work is to discuss how a non-paramet
ric approach based on the notion of the L^p data depth function can be exp
loited in the Statistical Process Control framework.\n\nhttps://conference
s.enbis.org/event/11/contributions/169/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/169/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Interactive tool for clustering and forecasting patterns of Taiwan
COVID-19 speared
DTSTART:20210914T152500Z
DTEND:20210914T154500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-142@conferences.enbis.org
DESCRIPTION:Speakers: Mahsa Ashouri\, Frederick Kin Hing Phoa (Academia Si
nica)\n\nThe COVID-19 data analysis is essential for policymakers in analy
zing the outbreak and managing the containment. Many approaches based on t
raditional time series clustering and forecasting methods such as hierarch
ical clustering and exponential smoothing have been proposed to cluster an
d forecast the COVID-19 data. However\, most of these methods do not scale
up with the high volume of cases. Moreover\, the interactive nature of th
e application demands further critically complex yet effective clustering
and forecasting techniques. In this paper\, we propose a web-based interac
tive tool to cluster and forecast the available data on Taiwan COVID-19 co
nfirmed infection cases. We apply the Model-based (MOB) tree and domain-re
levant attributes to cluster the dataset and display forecasting results u
sing the Ordinary Least Square (OLS) method. In this OLS model\, we apply
a model produced by the MOB tree to forecast all series in each cluster. O
ur user-friendly parametric forecasting method is computationally cheap. A
web app based on R's Shiny App makes it easier for the practitioners to f
ind clustering and forecasting results while choosing different parameters
such ad domain-relevant attributes. These results could help determine th
e spread pattern and be utilized by researchers in medical fields.\n\nhttp
s://conferences.enbis.org/event/11/contributions/142/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/142/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Two questions of "class": Kind of quantity and Classification
DTSTART:20210913T134500Z
DTEND:20210913T141500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-143@conferences.enbis.org
DESCRIPTION:Speakers: Jeanette Melin (RI.SE Metrology)\, Leslie Pendrill (
RI.SE Metrology)\n\nThe need to handle ordinal and nominal data is current
ly being addressed in various work going on amongst ontology organisations
and various standards bodies dealing with concept systems in response to
big data\, machine reading in applications such as the medical field. At t
he same time\, some prominent statisticians have been reticent about accep
ting someone else telling them what scales they should use when analysing
data. This presentation reviews how two key concepts - Kind of quantity an
d Classification - can be defined and form the basis for comparability\, a
dditivity\, dimensionality\, etc and are essential to include in any conce
pt system for Quantity. Examples include on-going research on neurodeneger
ation as studied in the European EMPIR project NeuroMET2.\n\nhttps://confe
rences.enbis.org/event/11/contributions/143/
LOCATION:Room 4
URL:https://conferences.enbis.org/event/11/contributions/143/
END:VEVENT
BEGIN:VEVENT
SUMMARY:An Ode to Tolerance: beyond the significance test and p-values
DTSTART:20210915T124000Z
DTEND:20210915T130000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-147@conferences.enbis.org
DESCRIPTION:Speakers: Olivier Cartiaux\, Walter Hoyer (GSK)\, Bernard Fran
cq\, Ron Kenett (KPA Group and Samuel Neaman Institute\, Technion\, Israel
)\, Dan Lin (GSK)\n\nIn comparative statistical tests of parallel treatmen
t groups\, a new drug is commonly considered superior to the current versi
on if the results are statistically significant. Significance is then base
d on confidence intervals and p-values\, the reporting of which is request
ed by most top-level medical journals. However\, in recent years there hav
e been ongoing debates on the usefulness of these parameters\, leading to
a ‘significance crisis’ in science.\n\nWe will show that this conventi
onal quest for statistical significance can lead to confusing and misleadi
ng conclusions for the patient\, as it focuses on the average difference b
etween treatment groups. By contrast\, prediction or tolerance intervals d
eliver information on the individual patient level\, and allow a clear int
erpretation following both frequentist and Bayesian paradigms. \n\nAdditio
nally\, treatment successes on the patient level can be compared using the
concept of individual superiority probability (ISP). While a p-value for
mean treatment effects converges to 0 or 1 when the sample size gets large
\, the ISP is shown to be independent of the sample size\, which constitut
es a major advantage over the conventional concept of statistical signific
ance. The relationship between p-values\, ISP\, confidence intervals and t
olerance intervals will be discussed and illustrated with analysis of some
real world data sets.\n\nhttps://conferences.enbis.org/event/11/contribut
ions/147/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/147/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Online Hierarchical Forecasting for Power Consumption Data
DTSTART:20210914T150500Z
DTEND:20210914T152500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-139@conferences.enbis.org
DESCRIPTION:Speakers: Malo Huard \, Margaux Brégère\n\nWe propose a thre
e-step approach to forecasting time series of electricity consumption at d
ifferent levels of household aggregation. These series are linked by hiera
rchical constraints -global consumption is the sum of regional consumption
\, for example. First\, benchmark forecasts are generated for all series u
sing generalized additive models\; second\, for each series\, the aggregat
ion algorithm `ML-Poly'\, introduced by Gaillard\, Stoltz and van Erven in
2014\, finds an optimal linear combination of the benchmarks\; Finally\,
the forecasts are projected onto a coherent subspace to ensure that the fi
nal forecasts satisfy the hierarchical constraints. By minimizing a regret
criterion\, we show that the aggregation and projection steps improve the
root mean square error of the forecasts. Our approach is tested on househ
old electricity consumption data\; experimental results suggest that succe
ssive aggregation and projection steps improve the benchmark forecasts at
different levels of household aggregation.results suggest that successive
aggregation and projection steps improve the benchmark forecasts at differ
ent levels of household aggregation. Results suggest that successive aggre
gation and projection steps improve the benchmark forecasts at different l
evels of household aggregation.\n\nhttps://conferences.enbis.org/event/11/
contributions/139/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/139/
END:VEVENT
BEGIN:VEVENT
SUMMARY:PHEBUS\, a Python package for the probabilistic seismic Hazard Est
imation through Bayesian Update of Source models
DTSTART:20210914T152500Z
DTEND:20210914T154500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-211@conferences.enbis.org
DESCRIPTION:Speakers: Jessie Mayor (EDF\, France)\, Clara Duverger (CEA\,
France)\, Gloria Senfaute (EDF R&D\, France)\, Merlin Keller (EDF R&D\, Fr
ance)\n\nWe propose a methodology for the selection and/or aggregation of
probabilistic seismic hazard analysis (PSHA) models\, which uses Bayes's t
heory by optimally exploiting all available observations\, in this case\,
the seismic and accelerometric databases. When compared to the actual meth
od of calculation\, the proposed approach\, simpler to implement\, allows
a significant reduction in computation time\, and more exhaustive use of t
he data.\nWe implement the proposed methodology to select the seismotecton
ic zoning model\, consisting of a subdivision of the national territory in
to regions that are assumed homogeneous in terms of seismicity\, amongst a
list of models proposed in the literature. Computation of Bayes factors a
llows comparing the adjustment performances of each model\, in relation to
a given seismic catalog. We provide a short description of the resulting
PHEBUS Python package structure and illustrate its application to the Fren
ch context.\n\nhttps://conferences.enbis.org/event/11/contributions/211/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/211/
END:VEVENT
BEGIN:VEVENT
SUMMARY:MODELLING WIND TURBINE POWER PRODUCTION WITH FUZZY LINEAR REGRESSI
ON METHODS
DTSTART:20210915T102000Z
DTEND:20210915T104000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-150@conferences.enbis.org
DESCRIPTION:Speakers: SADIK OZKAN GUNDUZ (HACETTEPE UNIVERSITY)\, OZLEM MU
GE TESTIK (HACETTEPE UNIVERSITY)\n\nWind energy is an immensely popular re
newable energy source\, due to the increase in environmental awareness\, t
he decrease in the number of fossil fuels\, and the increase in costs. The
refore\, the amount of energy produced in wind turbine farms should be est
imated accurately. Although wind turbine manufacturers estimate energy pro
duction depending on wind speed and wind direction\, mostly actual product
ions are different from these estimates. Such differences may be observed
not only because of model errors or randomness\, but also from uncertainty
in the environment\, or lack of data in the sample. In this study\, energ
y production is estimated by using wind speed and wind direction\, where e
ither measurement errors or vagueness mostly exist. In order to deal with
this disadvantage\, fuzzy logic is implemented in the proposed regression
models. Four different fuzzy regression models are constructed according t
o the fuzziness situation. Crisp (non-fuzzy) input crisp output\, crisp in
put fuzzy output\, and fuzzy input fuzzy output situations are considered\
, and the results are compared. Numerous fuzzy regression models are used
in this study and it is concluded that fuzzy models can both suggest effec
tive solutions where fuzziness exists\, and provide more flexible estimati
ons and decisions.\n\nhttps://conferences.enbis.org/event/11/contributions
/150/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/150/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Imbalanced multi-class classification in process industries. Case
study: Emission levels of SO2 from an industrial boiler
DTSTART:20210915T120000Z
DTEND:20210915T122000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-149@conferences.enbis.org
DESCRIPTION:Speakers: Gustavo Almeida (Federal University of Minas Gerais)
\, Tomás Carmo (Federal University of Minas Gerais)\n\nImbalanced classes
often occur in classification tasks including process industry applicatio
ns. This scenario usually results in the overfitting of the majority class
es. Imbalanced data techniques are then commonly used to overcome this iss
ue. They can be grouped into sampling procedures\, cost-sensitive strategi
es and ensemble learning. This work investigates some of them for the clas
sification of SO2 emissions from a kraft boiler belonging to a pulp mill i
n Brazil. There are six classes of emission levels\, where the available n
umber of samples of the highest one is considerably smaller since it refle
cts negative operating conditions. Four oversampling procedures\, namely S
MOTE\, ADASYN\, Borderline-SMOTE and Safe-level-SMOTE\, and the bagging (B
ootstrap Aggregating) ensemble method\, were investigated. All tests used
an MLP neural network with a single hidden layer. The number of hidden uni
ts ([1:1:16])\, the activation function (logistic\, hyperbolic tangent)\,
and the learning algorithm (Rprop\, LM\, BFGS)\, as well as the imbalance
ratio\, were also varied. The best results increased the AUC for the minor
ity class from 83.9% to 93.6%\, and from 80.4% to 89.1%\, which represents
a gain of about 10%\, while keeping the AUCs of the remaining classes pra
ctically unchanged. This significantly increased the individual g-mean met
ric for the minority class from 60.9% to 79.8%\, and from 52.9% to 76.3%\,
respectively\, without significant changes in the overall g-mean metric\,
as desired. All results are given in average values. Imbalanced multi-cla
ss data generally appear in process industries\, which claims the use of d
ata imbalanced strategies to achieve high accuracy for all classes.\n\nhtt
ps://conferences.enbis.org/event/11/contributions/149/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/149/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Parameter Calibration in wake effect simulation model with Stochas
tic Gradient Descent and stratified sampling
DTSTART:20210915T130000Z
DTEND:20210915T133000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-197@conferences.enbis.org
DESCRIPTION:Speakers: Eunshin Byon\, Bingjie Liu (University of Michigan)\
n\nAs the market share of wind energy has been rapidly growing\, wake effe
ct analysis is gaining substantial attention in the wind industry. Wake ef
fects represent a wind shade cast by upstream turbines to the downwind dir
ection\, resulting in power deficits in downstream turbines. To quantify t
he aggregated influence of wake effects on a wind farm's power generation\
, various simulation models have been developed\, including Jensen's wake
model. These models include parameters that need to be calibrated from fie
ld data. Existing calibration methods are based on surrogate models that i
mpute the data under the assumption that physical and/or computer trials a
re computationally expensive\, typically at the design stage. This\, howev
er\, is not the case where large volumes of data can be collected during t
he operational stage. Motivated by wind energy applications\, we develop a
new calibration approach for big data settings without the need for stati
stical emulators. Specifically\, we cast the problem into a stochastic opt
imization framework and employ stochastic gradient descent to iteratively
refine calibration parameters using randomly selected subsets of data. We
then propose a stratified sampling scheme that enables choosing more sampl
es from noisy and influential sampling regions and thus\, reducing the var
iance of the estimated gradient for improved convergence\n\nhttps://confer
ences.enbis.org/event/11/contributions/197/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/197/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Application of machine learning models to discriminate tourist lan
dscapes using eye-tracking data
DTSTART:20210914T130000Z
DTEND:20210914T133000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-168@conferences.enbis.org
DESCRIPTION:Speakers: Giulia Contu (University of Cagliari)\, Gianpaolo Za
mmarchi (University of Cagliari)\, Luca Frigau (University of Cagliari)\n\
nNowadays tourist websites make extensive use of images to promote their s
tructure and the its location. Many images\, such as landscapes\, are used
extensively on destination tourism websites to draw tourists’ interest
and influence their choices. The use of eye-tracking technology has improv
ed the level of knowledge of how different types of pictures are observed.
An eye-tracker enables to accurately define the eye location and therefor
e to carry out precise measurement of the eye movements during the visuali
zation of different stimuli (e.g. pictures\, documents). \nEye-tracking da
ta can be analyzed to convert the viewing behavior in terms of quantitativ
e measurements and they might be collected for a variety of purposes in a
variety of fields\, such as grouping clients\, improving the usability of
a website\, and in neuroscience studies. Our work aims to use eye-tracking
data from a publicly available repository to get insight of user behavior
regarding two main categories of images: natural landscapes and city land
scapes. We choose to analyze these data using supervised and unsupervised
methods. Finally\, we evaluate the results in terms of which choice should
be made between possible options to shed light on how decision-makers sho
uld take this information into account.\n\nhttps://conferences.enbis.org/e
vent/11/contributions/168/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/168/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Copula-based robust optimal block designs
DTSTART:20210914T140000Z
DTEND:20210914T143000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-170@conferences.enbis.org
DESCRIPTION:Speakers: Andreas Rappold\, Werner G. Mueller\, Dave Woods\n\n
Blocking is often used to reduce known variability in designed experiments
by collecting together homogeneous experimental units. A common modeling
assumption for such experiments is that responses from units within a bloc
k are dependent. Accounting for such dependencies in both the design of th
e experiment and the modeling of the resulting data when the response is n
ot normally distributed can be challenging\, particularly in terms of the
computation required to find an optimal design. The application of copulas
and marginal modeling provides a computationally efficient approach for e
stimating population-average treatment effects. Motivated by an experiment
from materials testing\, we develop and demonstrate designs with blocks o
f size two using copula models. Such designs are also important in applica
tions ranging from microarray experiments to experiments on human eyes or
limbs with naturally occurring blocks of size two. We present a methodolog
y for design selection\, make comparisons to existing approaches in the li
terature\, and assess the robustness of the designs to modeling assumption
s.\n\nhttps://conferences.enbis.org/event/11/contributions/170/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/170/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Generalized additive models for ensemble electricity demand foreca
sting
DTSTART:20210914T154500Z
DTEND:20210914T160500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-215@conferences.enbis.org
DESCRIPTION:Speakers: Simon N. Wood (University of Edinburgh)\, Yannig Gou
de (EDF R&D)\, Biagio Palumbo (University of Naples Federico II)\, Christi
an Capezza\, Matteo Fasiolo\n\nFuture grid management systems will coordin
ate distributed production and storage resources to manage\, in a cost-eff
ective fashion\,\nthe increased load and variability brought by the electr
ification of transportation and by a higher share of weather-dependent pro
duction.\nElectricity demand forecasts at a low level of aggregation will
be key inputs for such systems. In this talk\, I'll focus on forecasting d
emand at the individual household level\,\nwhich is more challenging than
forecasting aggregate demand\, due to the lower signal-to-noise ratio and
to the heterogeneity of consumption patterns across households.\nI'll desc
ribe a new ensemble method for probabilistic forecasting\, which borrows s
trength across the households while accommodating their individual idiosyn
crasies.\nThe first step consists of designing a set of models or 'experts
' which capture different demand dynamics and fitting each of them to the
data from each household.\nThen the idea is to construct an aggregation of
experts where the ensemble weights are estimated on the whole data set\,
the main innovation being that we let the weights vary with the covariates
by adopting an additive model structure. In particular\, the proposed agg
regation method is an extension of regression stacking (Breiman\, 1996) wh
ere the mixture weights are modelled using linear combinations of parametr
ic\, smooth or random effects.\nThe methods for building and fitting addit
ive stacking models are implemented by the gamFactory R package\, availabl
e at https://github.com/mfasiolo/gamFactory\n\nReferences:\n- Breiman\, L.
\, 1996. Stacked regressions. Machine learning\, 24(1)\, pp.49-64.\n\nhttp
s://conferences.enbis.org/event/11/contributions/215/
LOCATION:Room 1
URL:https://conferences.enbis.org/event/11/contributions/215/
END:VEVENT
BEGIN:VEVENT
SUMMARY:DATA MINING FOR DISCOVERING DEFECT ASSOCIATIONS AND PATTERNS TO IM
PROVE PRODUCT QUALITY: A CASE FOR PRINTED CIRCUIT BOARD ASSEMBLY
DTSTART:20210914T144500Z
DTEND:20210914T150500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-125@conferences.enbis.org
DESCRIPTION:Speakers: Ayse Merve Parlaktuna\, Murat Caner Testik (Hacettep
e University)\n\nMeeting customer quality expectations and delivering high
quality products is the key for operational excellence. In this study\, a
printed circuit board (PCB) assembly process is considered for improvemen
t. Associations between the defects as well as patterns of the defects ove
r time are investigated. A priori algorithm for association rule mining an
d Sequential Pattern Discovery using Equivalence classes (SPADE) algorithm
for pattern mining were implemented in R and SPMF\, respectively. A datas
et consisting of seven years of defect data standardized according to the
IPC Standard was prepared for this purpose. Association analysis was done
on the basis of card types and the years. It is concluded that association
s between defect types change according to the card type due to design par
ameters. Pattern analysis indicated that some defect types are recurring o
ver time. For example\, insufficient solder and tombstone defect types rec
urred over and over. On the other hand\, there were also some defect types
\, such as excess solder defects causing solder balls\, that occurred sequ
entially. As the root causes of excess solder defects were eliminated\, mo
st of the potential solder ball defects were also eliminated. In the follo
wing\, preparation of the dataset for analyses\, implementation\, and resu
lts of the study are discussed with examples.\n\nhttps://conferences.enbis
.org/event/11/contributions/125/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/125/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Statistical models for measurement uncertainty evaluation in coord
inate metrology
DTSTART:20210915T133000Z
DTEND:20210915T140000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-200@conferences.enbis.org
DESCRIPTION:Speakers: Alistair Forbes (National Physical Laboratory)\n\nCo
ordinate metrology is a key technology supporting the quality infrastructu
re associated with manufacturing. Coordinate metrology can be thought of a
s a two-stage process\, the first stage using a coordinate measuring machi
ne (CMM) to gather coordinate data $\\mathbf{x}_{1:m} = \\{\\mathbf{x}_i\
,i =1\, \\ldots\,m\\}$\, related to a workpiece surface\, the second extra
cting a set of parameters (features\, characteristics) $\\mathbf{a} = (a_
1\,\\ldots\,a_n)^\\top$ from the data e.g.\, determining the parameters a
ssociated with the best-fit cylinder to data. The extracted parameters can
then be compared with the workpiece design to assess whether or not the m
anufactured workpiece conforms to design within prespecified tolerance.\n\
nThe evaluation of the uncertainties associated with geometric features $\
\mathbf{a}$ derived from coordinate data $\\mathbf{x}_{1:m}$ is also a tw
o stage process\, the first in which a $3m \\times 3m$ variance matrix $V_
X$ associated with the coordinate data is evaluated\, the second stage in
which these variances are propagated through to those for the features $\\
mathbf{a}$ derived from $\\mathbf{x}_{1:m}$. While the true variance matr
ix associated with a point cloud may be difficult to evaluate\, a reasonab
le estimate can be determined using approximate models of CMM behaviour. \
nIn this paper we describe approximate models of CMM behaviour in terms of
spatial correlation models operating at different length scales and show
how the point cloud variance matrix generated using these approximate mode
ls can be propagated through to derived features. We also use the models t
o derive explicit formulae that characterise the uncertainties associated
with commonly derived parameters such as the radius of a fitted cylinder.\
n\nhttps://conferences.enbis.org/event/11/contributions/200/
LOCATION:Room 2
URL:https://conferences.enbis.org/event/11/contributions/200/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Deciphering Random Forest models through conditional variable impo
rtance
DTSTART:20210914T094000Z
DTEND:20210914T100000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-164@conferences.enbis.org
DESCRIPTION:Speakers: Marta Rotari\, murat kulahci\n\nIn many data analyti
cs applications based on machine learning algorithms\, the main focus is u
sually on predictive modeling. In certain cases\, as in many applications
in manufacturing\, understanding the data-driven model plays a crucial rol
e in complementing the engineering knowledge about the production process.
There is therefore a growing interest in describing the contributions of
the input variables to the model in the form of “variable importance”\
, which is readily available in certain machine learning methods such as r
andom forest (RF). In this study\, we focus on the Boruta algorithm\, whic
h is an effective tool in determining the importance of variables in RF mo
dels. In many industrial applications with multiple input variables\, it b
ecomes likely to observe high correlation among these variables. It is sho
wn that the correlation among the input variables distorts and overestimat
es the importance of variables. The Boruta algorithm is also affected by t
his resulting in a larger set of input variables deemed important. To over
come this\, in this study we present an extension of the Boruta algorithm
for the correlated data by exploiting the conditional importance\, which t
akes into consideration the correlation structure of the variables for com
puting the importance scores. This leads to a significant improvement of t
he variable importance scores in the case of a high correlation among vari
ables and to a more precise ranking of the variables that contribute to th
e model significantly. We believe this approach can be used in many indust
rial applications by providing more transparency and understanding of the
process.\n\nhttps://conferences.enbis.org/event/11/contributions/164/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/164/
END:VEVENT
BEGIN:VEVENT
SUMMARY:IMPORTANCE OF SPATIAL DEPENDENCE IN THE CLUSTERING OF NDVI FUNCTIO
NAL DATA ACROSS THE ECUADORIAN ANDES
DTSTART:20210915T124000Z
DTEND:20210915T130000Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-212@conferences.enbis.org
DESCRIPTION:Speakers: Jorge Mateu (Universidad Jaume I)\, Miguel Flores (M
ODES\, SIGTIG\,Escuela Politécnica Nacional)\, Xavier Zapata-Ríos (Escue
la Politécnica Nacional)\, Jeysson Chuquin (Escuela Politécnica Nacional
)\, Sandra Torres (Dirección de Estudios e Investigación\, INAMHI)\, Ale
xandra Maigua (Escuela Politécnica Nacional)\n\nThe spatial dependence on
environmental data is an influential criterion in clustering processes\,
since the results obtained provide relevant information. As classical meth
ods do not consider spatial dependence\, considering this structure produc
es unexpected results\, and groupings of curves that cannot be similar in
shape/behavior.\nIn this work\, the clustering is performed using the modi
fied k-means method for spatially correlated functional data applied to ND
VI data from the ecuadorian Andes. NDVI studies are important because it i
s used mainly to measure biomass\, assess crop health\, help forecast fire
danger zones\, etc.\nFor this\, quality indexes are implemented that can
obtain the appropriate number of groups. Based on the methodology used in
the hierarchical approach for functional data with spatial correlation\, a
nd given that the functional data belong to the Hilbert space of square-in
tegrable functions\; the analysis is developed considering the distance be
tween curves through the $\\mathcal{L}^2$ norm\, obtaining a reduced repre
sentation of the data through a finite Fourier-type basis. Then\, the empi
rical variogram is calculated and a parametric theoretical model is fitted
in order to weight the distance matrix between the curves by the trace-va
riogram and multivariogram calculated with the coefficients of the base fu
nctions\, this matrix carries out the grouping of spatially correlated fun
ctional data. For the validation of the method\, some simulation scenarios
were carried out\, obtaining more than $80 \\%$ of good classification an
d complemented with a case of application to NDVI data\; obtaining five la
titudinally distributed regions\; these regions are influenced by the hydr
ographic basins of Ecuador.\n\nhttps://conferences.enbis.org/event/11/cont
ributions/212/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/212/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Spatial correction of low-cost sensors observations for fusion of
air quality measurements
DTSTART:20210914T150500Z
DTEND:20210914T152500Z
DTSTAMP:20240303T015300Z
UID:indico-contribution-162@conferences.enbis.org
DESCRIPTION:Speakers: Michel Bobbia (Atmo Normandie)\, Bruno Portier (INSA
Rouen Normandie)\, Jean-Michel Poggi (University Paris-Saclay)\n\nThe con
text is the statistical fusion of air quality measurements coming from dif
ferent monitoring networks. The first one of fixed sensors of high quality
\, the reference network\, and the second one of micro-sensors of less qua
lity. Pollution maps are obtained from the correction of numerical model o
utputs using the measurements from the monitoring stations of air quality
networks. Increasing the density of sensors would then improve the quality
of the reconstructed map. The recent availability of low-cost sensors in
addition to reference station measurements makes it possible without prohi
bitive cost. \nUsually\, a geostatistical approach is used for the fusion
of measurements but the first step is to correct micro-sensors measures th
anks to those given by the reference sensors by prior offline fitting a mo
del issued from a costly and sometimes impossible colocation period. We pr
opose to complement these approaches by considering online spatial correct
ion of micro-sensors performed simultaneously with data fusion. The basic
idea is to use the reference network to correct the measures from network
2: the reference measurements are first estimated by kriging only the meas
urements of network 2\; then the residuals of the estimation on network 1
are calculated\; and finally\, the correction to be applied to the micro-s
ensors is obtained by kriging these residuals. Then we can iterate or not
this sequence of steps\, and alternate or not the role of the networks dur
ing the iterations. \nThis algorithm is first introduced\, then explored b
y simulation\, and then applied to a real-world dataset.\n\nhttps://confer
ences.enbis.org/event/11/contributions/162/
LOCATION:Room 3
URL:https://conferences.enbis.org/event/11/contributions/162/
END:VEVENT
END:VCALENDAR