Oslo section of the Norwegian Statistical Association
Latest News
📰 New issue of Tilfeldig gang
A new issue of TG (no. 1, vol. 42) is now available on the website (in Norwegian)!
🎊 New board for the Oslo section of the Norwegian Statistical Association!
Head: Clara Bertinelli Salucci, UiO
Treasurer: Lars Henry Berge Olsen, NR
Secretary: Simon Lergenmuller, FHI
Website manager: Zhi Zhao, UiO
Board member: Marthe Elisabeth Aastveit, UiO/NR
Board member: Vera Haugen Kvisgaard, UiO
Statistics evening and member meetings
🎉 22.05.2025
“Statistical modelling using gradient boosting algorithms: applications across diverse fields” with Riccardo De Bin
Program:
17.00 - 17.20: Pizza, sushi and beverages
17.20 - 17.30: Annual meeting
17.30 - 18.30: Presentation by Riccardo De Bin
Place: Norwegian Computing Center (NR), (Alfa-Omega room, Gaustadalléen 23A, 0373 Oslo)
Speaker: Riccardo De Bin, UiO
Title: Statistical modelling using gradient boosting algorithms: applications across diverse fields
Abstract:
In this talk we will see how the gradient boosting approach can be used to fit statistical models in various contexts: in the analysis of omics data through First Hitting Time models; in the prediction of the remaining useful life of lithium-ion batteries; in the assessment of airport runway conditions for safe landings; and in the analysis of the properties of transition metal complexes for catalysis. Statistical issues addressed in this talk include the extensions of classical survival models to the high dimensional frameworks, where the number of variables is (much) larger than the number of observations; the handling of informative censoring in survival models; and the exploration of graph structures.
🎉 16.01.2025
“Statistics is biology’s next microscope. Biostatistical research collaborations interpreted in knitted designs” with Kathrine Frey Frøslie
Program:
17.00 - 17.30: Pizza, sushi and beverages
17.30 - 18.30: Presentation by Kathrine Frey Frøslie
Place: NMBU, Ås, Library at biotechnology building (Chr. Magnus Falsens v. 16, 1433 Ås) and Teams
Speaker: Riccardo De Bin, UiO
Title: Statistics is biology’s next microscope. Biostatistical research collaborations interpreted in knitted designs
Abstract:
The talk will be a fusion of art and science. You will both get an overview of the research topics, the design processes, and see the artwork IRL.
🎉 xx.xx.2024
…
🎉 28.11.2023
“Modellering av en forskers publiserings- og siteringsrate” med Magne Aldrin
Program:
17.00 - 17.30: Pizza, sushi og drikke
17.30 - 18.30: Presentasjon ved Magne Aldrin
Hvor: Universitetet i Oslo, (Abels utsikt, 12. etasje, Niels Henrik Abels hus - Universitetet I Oslo, Moltke Moes vei 35) og Zoom
Foredragsholder: Magne Aldrin, Forskningsleder, NR
Tittel: Modellering av en forskers publiserings- og siteringsrate
Abstrakt:
Blir forskere mer produktive når de blir eldre? Er forskeren som tok doktorgrad tidlig i sin karriere mer produktiv enn de som gjorde det seinere? Vi, dvs. min kollega Olav Nikolai Breivik og jeg, undersøker dette ved å modellere her hvordan publiseringsraten (antall artikler o.l. per år) og siteringsraten til en gitt forsker forskjellige over tid som en funksjon av forklaringsvariabler knyttet til forskeren. Av særskilt interesse er om en forsker som får tildelt et såkalt FRIPRO-prosjektet fra Norges forskningsråd (NFR) får økt sin vitenskapelige produksjon sammenlignet med en sammenlignbar forsker som ikke får et slikt prosjekt. For å undersøke dette har vi et brukt et datasett med 2500 forskere som til sammen sendte inn 5000 søknader om FRIPRO-prosjekter i perioden 2004-2012, med informasjon om deres publiseringshistorie før og etter søknaden, til sammen 70000 publikasjoner. Vi modellerer årlig publiserings- og siteringsrate i etterkant av søknaden som funksjon av blant annet om forskeren fikk innvilget søknaden eller ikke, tid etter søknaden, alder, kjønn og tidligere publikasjonshistorie. Vi bruker generaliserte additive modeller (GAM). Modellene er programmert i programmeringssystemet TMB, som gir mulighet for fleksibel modellering og rask optimering basert på automatisk avledning. Dette systemet finnes nå i en rein R-versjon, RTMB, og jeg vil gi en kort innføring til det som en liten smakebit til TMB-kurset som Olav Nikolai Breivik vil holde i forkant av neste års norske statistikkmøte i Tønsberg. Arbeidet som presenteres har blitt utført på oppdrag fra Norges forskningsråd.
🎉 09.05.2023
“Bayes, Confidence Distributions, and US Pre-War Economy” with Nils Lid Hjort
Program:
17.00 - 17.30: Pizza, sushi and beverages
17.30 - 18.30: Presentation by Nils Lid Hjort
Place: Abels utsikt (12th floor, Niels Henrik Abels hus - Universitetet i Oslo) and Zoom
Speaker: Nils Lid Hjort, UiO
Title: Bayes, Confidence Distributions, and US Pre-War Economy
Abstract:
Confidence distributions are frequentist analogues of the Bayesian posterior distributions, but without priors. Often these two approaches lead to similar results, regarding post-data intervals for primary parameters, etc. Situations with boundary parameters are different and more complicated, however. If your focus parameter is known to be nonnegative, a common Bayesian solution is to work with a flat prior from zero to infinity. I show that this might lead to very different and ostensibly wrong results, compared to confidence distributions.
When Christopher Sims won the Sveriges Riksbank Prize in Economic Sciences im Memory of Alfred Nobel for 2011 (the so-called Nobel Prize in Economics), he chose for his Stockholm prize lecture to showcase how Bayesian modelling with MCMC worked for the analysis of a certain dataset on pre-war US macroeconomics (involving consumption, investment, government spending, 1929 to 1940). I illustrate the apparatus of confidence distributions with t-bootstrapping to analyse the same data and the same vector autoregressive time series model, but reaching rather different results when assessing the main parameters.
The essence of this story is contained in my book Confidence, Likelihood, Probability, with Tore Schweder, but I have more material and a different presentation of the main reasons for the differences with the Bayesian setup.
🎉 30.11.2022
“The Brier score for evaluating time-to-event predictions for administratively censored data: problems and a solution” with Ørnulf Borgan
Program:
17.00 - 17.30: Pizza, sushi and beverages
17.30 - 18.30: Presentation by Ørnulf Borgan
Place: Abels utsikt (12th floor, Niels Henrik Abels hus - Universitetet i Oslo) and Zoom
Speaker: Ørnulf Borgan, UiO
Title: The Brier score for evaluating time-to-event predictions for administratively censored data: problems and a solution
Abstract:
The Brier score is commonly used for evaluating probability predictions. For time-to-event predictions based on right-censored observations of the event times, the score can be weighted by the inverse probability of censoring (IPCW) to retain its original interpretation. In the talk I will consider situations where individuals are recruited to a study population at different calendar times, and then followed up to the occurrence of an event of interest or to the closure of the study at a given date. For such administratively censored data, there may be problems with the IPCW weighting scheme when the distribution of the covariates varies with calendar time. I will discuss the reason for these problems, and also suggest an alternative version of the Brier score that does not suffer from the same problems. The talk is based on a joint paper with Håvard Kvamme that will soon appear in the Journal of Machine Learning Research.
Local institutions
