Estimating the Effect of Second Screening on TV Viewing using Mobile Diaries

The Luxembourg Statistical Society would like to invite you to the talk by Prof. Sarah Gelper on “Estimating the Effect of Second Screening on TV Viewing using Mobile Diaries”. The talk will take place on Thursday 23rd February at 12:30 PM in hybrid mode. If you attend it in person, it will be in the room on the first floor of the Maison du Nombre, MNO-E01-0146030, on the University Campus Belval.

Sarah Gelper is Associate Professor in Business Analytics with a Specialization in Marketing Analytics at Université du Luxembourg.


People often use second devices, such as their mobile or tablet, while watching TV. Such « second screening » might enhance engagement and increase viewership but also risks distracting viewers and reducing interest in the show. Estimating the effect of second screening on repeat viewing is challenging for three reasons. (1) We need to observe behavior on two different devices (the data challenge); (2) the effect of second screening might vary with viewer and show characteristics (the heterogeneity challenge); and (3) second screening is highly endogenous behavior (the endogeneity challenge). This paper leverages large-scale individual-level data from mobile diaries of 1,702 US TV viewers on 2,755 primetime TV shows over 3-6 weeks in which viewers report both viewing and second screening. The data are augmented to account for a broad set of marketing influences such as show communications and promos. We characterize heterogeneity by four latent viewer classes (heavy viewers, sports enthusiasts, sitcom viewers, and light viewers) and use a causal forest analysis to estimate the heterogeneous effects of second screening on repeat viewing. Despite the possible risk of distraction, second screening does not harm actual TV viewing, and this effect is robust over various types of second screening activities and across viewer segments. Second screening impacts attitudes more than behaviors – unlike actual viewing, stated viewing intentions are affected right after the second screening experience. Specifically, viewing intentions increase when people use a second screen to communicate about the show, but decrease when they use the second screen for show-unrelated activities.

Practical information

Date: 23 February 2023
Time: 13:00-14:00
Place: room MNO-E01-0146030 (Belval campus)
Fee: free
Registration: please register by sending an email to Christophe Ley (Société Statistique Luxembourgeoise)

Random Forests: Introduction and industrial applications

The Luxembourg Statistical Society would like to invite you to the talk by Prof. Jean-Michel Poggi on “Random Forests: Introduction and industrial applications”. The talk will take place on Thursday 27th October at 1:00 PM and will be hosted in room MNO 1.040 (Belval campus).

Jean-Michel Poggi is Professor of Statistics at Université Paris Cité and member of the Lab. Maths Orsay (LMO) at Université Paris-Saclay, in France.
His research interests are in nonparametric time series, wavelets, tree-based methods (CART, Random Forests, Boosting) and applied statistics. Research activities combine theoretical and practical contributions together with industrial applications (mainly environment and energy) and software development.
He is Associate Editor of three journals: Journal of Statistical Software (JSS), Advances in Data Analysis and Classification (ADAC) and Journal of Data Science, Statistics, and Visualisation (JDSSV).
He is President of the European Network for Business and Industrial Statistics (ENBIS). He was President of the French Statistical Society (SFdS) from 2011 to 2013, and President of the European Courses in Advanced Statistics (ECAS) from 2015 to 2017.


Random forests (RF) are a statistical learning method extensively used in many fields of application, thanks to its excellent predictive performance. RF are part of the family of tree-based methods and inherit intrinsic flexibility of trees: adapted to both classification and regression problems, easily extended to many different types of data.
We will first focus on an introduction of RF, the definition of the variable importance measures and the related variable selection capability. We then illustrate their practical power in two different applied contexts. The first one relates to physiological signal processing and addresses the functional variable selection for driver’s stress level classification. The second is about the aggregation of multi-scale experts for bottom-up electricity load forecasting.

Practical information

Date: 27 October 2022
Time: 13:00-14:00
Place: room MNO 1.040 (Belval campus)
Fee: free
Registration: please register by sending an email to Christophe Ley (Société Statistique Luxembourgeoise)

22 octobre : Journée mondiale de la statistique

À l’occasion de la Journée mondiale de la statistique, nous vous invitons à découvrir comment l’intelligence artificielle pénètre dans des domaines aussi variés que les diagnostics médicaux, la physique des particules ou encore l’astronomie.

La leçon inaugurale sera donnée par le président de la Société Statistique Luxembourgeoise et professeur en statistique appliquée à l’,
Dr Christophe Ley.

Horaire : Samedi 12 octobre 2022 de 14h00 à 18h00 au Luxembourg Science Center

Adresse: Luxembourg Science Center – 1 rue John E. Dolibois, L-4620 Differdange

Attention: places limitées!

Inscriptions : Pour participer, veuillez envoyer un email à

In search of a universal estimator

Luxembourg Statistical Society has the pleasure to invite you to our second event of our society this year – an hybrid talk by Prof. Yannick Baraud (Department of Mathematics, Université du Luxembourg, Luxembourg).


In this talk, Prof. Yannick Baraud will present some of the research topics of the SanDAL* group. One of them is the search of a universal estimation procedure, i.e. a procedure which is applicable to all statistical models and automatically leads to an estimator of the distribution of the data that is both optimal and stable to the presence of outliers among the sample. We shall see how mathematics allow us to analyze the non-asymptotic properties of this new type of estimator, i.e. when the number of data is fixed and not necessarily large. Finally, we will illustrate its performance on some simulations by comparing it to some classical estimators.

*SanDAL is a Chair in Mathematical Statistics and Data Science at the University of Luxembourg which is funded by the European Union’s Horizon 2020 programme.

Practical information

Date: 28 March 2022
Time: 18:00 – 19:00
Place: online and in person
Fee: free
Registration: mandatory at Luxstat

COVID by numbers

Luxembourg Statistical Society has the pleasure to invite you to a truly extraordinary first event of our society this year – an online talk of Sir Prof. David Spiegelhalter (Statistical Laboratory, Centre for Mathematical Sciences, University of Cambridge, UK) and Dr Anthony Masters, Statistical Ambassador for the Royal Statistical Society.

Prof. Spiegelhalter is a renowned British statistician, currently Chair of the Winton Centre for Risk and Evidence Communication in the Department of Pure Mathematics and Mathematical Statistics at the University of Cambridge.

Dr Masters as a Chartered Statistician and a Statistical Ambassador for the Royal Statistical Society is widely involved in explaining statistical concepts among the general public. His column in The Guardian about data and the Covid-19 pandemic gained large popularity.


In his talk, Prof. Spiegelhalter and Dr Masters will share insights related to Covid-19 statistics and some key findings from their new book “Covid by Numbers: Making Sense of the Pandemic with Data” .

Practical information

Date: 16 February 2022
Time: 12:00 – 13:00
Place: online
Fee: free
Registration: mandatory at Agnieszka

New developments in high dimensional statistics

Luxembourg Statistical Society would like to promote the talks by Mark Podolskij on “New developments in high dimensional statistics” organised by the University of Luxembourg. This talks will take place on Tuesday 8th February at 11:00 AM and will be organized at Maison du Savoir, Room 3.370, Belval Campus.

Mark Podolskij is a full professor in financial mathematics at the departments of mathematics and finance. He has obtained his PhD in 2006 at the University of Bochum and held professor positions at the University of Heidelberg (2010-2014) and University of Aarhus (2014-2020). His research is focusing on statistical and probabilistic analysis of stochastic processes with applications to finance and physics. He currently holds an ERC Consolidator Grant « Statistical methods for high dimensional diffusions »..


In his talk, Mark Podolskij will present the challenges in high dimensional statistics and some new developments in the field. After the presentation of the main concepts and mathematical problems, he will focus on recent high dimensional theory for stochastic processes. Applications in finance will be discussed

Practical information

Date: 8 February 2022
Time: 11:00
Place: Maison du savoir, room 3.370, Belval Campus
Fee: free
Registration: mandatory at

Design and analysis of COVID vaccine trials : the familiar and the strange

The Luxembourg Statistical Society would like to promote the talks by Stephen Seen and Dr. Holly Jane on Design and analysis of COVID vaccine trials : the familiar and the strange organised by the Société Française de Statistique (SFdS). These talks will take place on Friday 3rd December at 5:00 PM and will be hosted online.

Dr. Holly Jane is a biostatistician working on the design and analysis of vaccine studies, with a particular expertise in HIV and COVID prevention and vaccine science. She is co-PI of the Statisics and Data Management Center of the HIV Vaccine Trials Network and a key investigator in the newly-formed USG-funded COVID Prevention Network.

Stephen Senn is consultant Statistician, worked in the pharmaceutical industry, in the National Health Service, further education and medical research, specialized in statistical methods for designing and analysing drug development programmes and clinical trials


Stephen Senn will start by giving a general presentation on the trials performed to evaluate COVID19 vaccines, discussing some choices such as sequential approach, 2:1 allocation, frequentist vs Bayesian.

Then Dr. Holly Janes will go deeper on some aspects such as choosing appropriate endpoints and long-term-assessment of vaccine efficacy with emergence of new variants.

Practical information

Date: 3 December 2021
Time: 17:00
Place: online
Fee: free
Registration: please register on SFdS website

Maternal depression and child human capital: an instrumental approach

The Luxembourg Statistical Society would like to invite you to the talk by Prof. Conchita d’Ambrosio on Maternal depression and child human capital: an instrumental approach. The talk will take place on Friday 12th November at 12:30 PM and will be hosted online.

Prof. Conchita d’Ambrosio has been Professor of Economics at the University of Luxembourg since 2013. Before joining the University of Luxembourg, Prof. D’Ambrosio was Associate Professor of Economics at the University of Milano-Biococca. Her research mainly focuses on income and wealth distributions, deprivation, polarization and social exclusion


In her talk Prof. D’Ambrosio will address the causal relationship between maternal depression and child human capital using UK cohort data. The study exploits conditionally exogenous variation in mothers’ genomes in an instrumental-variable approach and describes the conditions under which mother’s genetic variants can be used as valid instruments.

Practical information

Date: 12 November 2021
Time: 12:30-13:30
Place: online
Fee: free
Registration: please register by sending an email to Agnieszka Walczak (

Comme par hasART

Un évènement insolite organisé à l’occasion de la Journée Européenne de la Statistique.

Au programme:

– Kick-off de la 1ère édition de la European Statistics Competition au Luxembourg.

– Conférence grand public sur l’art du hasard par Yvik Swan, professeur de Mathématique à l’Université libre de Bruxelles, président de la Belgian Mathematical Society.

– Inauguration d’« Exploratis », une nouvelle station expérimentale au Science Center basé sur le célèbre livre interactif « Mathema » de Hugo Parlier et Paul Turner.

Informations pratiques

Date: 23 Octobre 2021
Time: 15:00-18:00
Place: Luxembourg Science Center (1, rue John E. Dolibois, Differdange)
Fee: gratuit

Doing Some Good with Machine Learning

It is a pleasure to invite you to our third LSS Presentation of 2021, to take place on June 4, from 8pm – 9:30 pm via Microsoft Teams.

The talk will be given by Prof. Lester Mackey. Lester Mackey is a statistical machine learning researcher at Microsoft Research New England and an adjunct professor at Stanford University. He received his Ph.D. in Computer Science (2012) and my M.A. in Statistics (2011) from UC Berkeley and my B.S.E. in Computer Science (2007) from Princeton University. Before joining Microsoft, he spent three years as an assistant professor of Statistics and, by courtesy, Computer Science at Stanford and one as a Simons Math+X postdoctoral fellow, working with Emmanuel Candes. His Ph.D. advisor was Mike Jordan, and his undergraduate research advisors were Maria Klawe and David Walker. He got his first taste of research at the Research Science Institute and learned to think deeply of simple things at the Ross Program


This is the story of my assorted attempts to do some good with machine learning. Through its telling, I’ll highlight several models of organizing social good efforts, describe half a dozen social good problems that would benefit from our community’s attention, and present both resources and challenges for those looking to do some good with ML.

Practical information

Date: 4 June 2021
Time: 20:00-21:30
Place: Microsoft Teams
Fee: Free
Registration: please register by sending an email to Michela Bia (