Trondheim Symposium in Statistics 2022

Back to the overview

The symposium took place on Friday 28 October – Saturday 29 October 2022 at Bårdshaug Herregård.

Invited speakers

• Magne Aldrin, Research Leader, Norwegian Computing Center, Oslo
• Thorsten Dickhaus, Professor of Mathematical Statistics, University of Bremen, Germany
• Geir Egil Eide , Professor Emeritus, University of Bergen, and at Helse Vest
• Ingrid Måge, Senior Scientist, Nofima, Ås
• Valeria Vitelli, Associate Professor, Unversity of Oslo

Organising committee: Øyvind Bakke and Mette Langaas.

14:15  Bus from NTNU Hovedbygningen to Bårdshaug Herregård
15:00  Check-in and coffee/cake
16:00–16:50  Thorsten Dickhaus
17:00–17:50  Geir Egil Eide
19:00  Dinner! After dinner: Jaktbaren

–9:00  Breakfast
9:00–9:50  Valeria Vitelli
10:00–10:50  Ingrid Måge
11:00–11:50  Magne Aldrin
12:00  Lunch
13:00  Bus to NTNU Gløshaugen

Time for presentations includes time for questions.

Thorsten Dickhaus: University of Bremen, Institute for Statistics
Title: Multiple testing of partial conjunction null hypotheses with application to replicability analysis of high-dimensional studies
Abstract: The partial conjunction null hypothesis is tested in order to discover a signal that is present in multiple studies. The standard approach of carrying out a multiple test procedure on the partial conjunction (PC) p-values can be extremely conservative. We suggest alleviating this conservativeness, by eliminating many of the conservative PC p-values prior to the application of a multiple test procedure. This leads to the following two-stage procedure: First, select the set with PC p-values below a selection threshold; second, within the selected set only, apply a family-wise error rate or false discovery rate (FDR) controlling procedure on the conditional PC p-values. We prove that the conditional PC p-values are valid for certain classes of one-parametric statistical models (including one-parameter natural exponential families), and we provide conditions for (asymptotic) FDR control for several multiple test procedures operating on conditional PC p-values. We also compare the proposed methodology with other recent approaches. This is joint work with Ruth Heller and Anh-Tuan Hoang.

Geir Egil Eide: UiB and Helse-Vest
Title: Attributable risk in a historical perspective with personal annotations; the past, the present and the future?
Abstract: Recent news have echoed results from the 2019 Global Burden of Disease Study about how many cancer deaths globally can be attributed to behavioural, environmental and occupational, and metabolic risk factors. Central in these estimates is the epidemiologic concept of attributable fraction or risk. I will give a short historic perspective on the conceptual development from the single factor to multifactorial setting, with personal notes, and describe their current use and some recent developments. This includes a shift in paradigm from automated regression analysis to models based on causal reasoning. Graphical illustrations and examples from respiratory disease epidemiology accompany the presentation.

Valeria Vitelli: Institute of Basic Medical Sciences, Faculty of Medicine, University of Oslo
Title: The Bayesian Mallows model and its recent developments – from preference learning to rank-based transcriptomic analyses
Abstract: The use of ranks in genomics is naturally linked to the underlying biological question, since one is often interested in overly-expressed genes in a given pathology. I propose to use a Bayesian Mallows model for ranks, able to both produce estimates of the consensus ranking of the genes shared among samples, and to fill-in missing data information. Interestingly, the model has already been fruitfully applied in other contexts, such as recommender systems, where it has proved to be useful for learning individualized preferences of the users, useful for providing personalized suggestions. Both when used in the context of genomics studies, and in user-oriented applications, posterior distributions of the unknowns are particularly relevant, since these can provide an evaluation of the uncertainty associated to the estimates, and thus of the reliability of the results. I will briefly review some relevant case studies to show the method's potentialities in the variety of situations in which we applied it, from genomic data integration to recommender systems. I will then conclude with a brief teaser on the most recent advancements and extensions.

Ingrid Måge: Nofima
Title: Statistics in food industry – applications and challenges
Abstract: Variability is inherent in all production processes, and the aim of industrial statistics is to improve products and processes by analysing, monitoring, and controlling this inherent variation. One of the biggest challenges for the food industry is variation in raw material quality, which again leads to variation in the final product. It is therefore important to be able to measure food quality, to characterize the variation, understand causes of variation, and develop strategies to control or reduce variation in the end-product. I will present Nofima’s ongoing research within these topics, with examples from some real food industry processes. The presented research is part of SFI DigiFoods (www.digifoods.no), a Centre for Research-based Innovation founded by the Norwegian Research Council.

Magne Aldrin: Norsk Regnesentral
Title: Estimation of climate sensitivity based on a simple climate model fitted to observations of hemispheric temperatures and global ocean heat content
Abstract: Predictions of climate change are uncertain mainly because of uncertainties in the emissions of greenhouse gases and how sensitive the climate is to changes in the abundance of the atmospheric constituents. The equilibrium climate sensitivity is defined as the temperature increase because of a doubling of the CO2 concentration in the atmosphere when the climate reaches a new steady state. CO2 is only one out of the several external factors that affect the global temperature, called radiative forcing mechanisms as a collective term. In this talk, I present a model framework for estimating the climate sensitivity. The core of the model is a simple, deterministic climate model based on elementary physical laws such as energy balance. It models yearly hemispheric surface temperature and global ocean heat content as a function of historical radiative forcing. This deterministic model is combined with an empirical, stochastic model and fitted to observations on hemispheric temperature and ocean heat content above and below 700 m, conditioned on estimates of historical radiative forcing. We use a Bayesian framework, with informative priors on a subset of the parameters and flat priors on the climate sensitivity and the remaining parameters. The model is estimated by Markov Chain Monte Carlo techniques.