Statistical Modelling Society

#IWSM2025 has officially kicked off with an amazingly insightful short course on physics-informed statistical learning today! If you don't want to miss any news, make sure to follow us on bluesky (statmodsociety.bsky.social) and LinkedIn (Statistical Modelling Society)!

We are now also on Bluesky @statmodsociety.bsky.social ! 🦋 We invite everyone to follow us on Bluesky and LinkedIn to stay updated with the latest news on #IWSM2025 and #SMS 🔔

☘️ #IWSM2025 in Limerick is approaching and we can't wait! iwsm2025.ie 📝 The deadline for papers has just been extended until February 24th. We are looking forward to your excellent work in statistical modelling! 📅 Early-bird registration will open soon!

#IWSM2024 is officially over 🥲 💬📃In total there were 55 inspiring contributed talks, 36 posters and 5 invited talks. 🥳🍀We hoped you enjoyed the workshop as much as we did and we can't wait to see all the old and new faces next July (13th-18th) in Limerick!

Finally, Kevin Burke gives the very last presentation of #IWSM2024. He sheds light on how to model social interaction data. They were particularly interested in reciprocation and ingroup effects. An agent-based null model was simulated to assess differences to random games.

Rui Martins gives insights into how he ended up in sports statistics and derives dynamical football predictions via log-normal priors for proportions. ⚽

The very last session of this year's #IWSM2024 is opened by Joaquin Cavieres who talks about how to use MCMC for relative abundance calculations. His solution is based on approximated Gaussian random fields under different parameterizations.

The last invited talk at #IWSM2024 is given by @ernstwit72 and deals with modelling dynamic networks. This marvelous presentation provides a new perspective on networks by moving away from unrealistic static graphs to longitudinal relational processes and their extensios. 🌐💡

The session is closed by Oskar Laverny who uses non-parametric approaches to estimate net survival under dependence between death causes. In particular, he estimates the excess survival under incomplete information about death causes.

Next up, Martin Boer talks about residual maximum likelihood for two-dimensional P-splines. His presentation includes illustrative examples on how to use the R package LMMsolver.

The last day starts with a presentation by @mbcarvalho about semiparametric Bayesian modeling of nonstationary joint extremes. He introduces an extremal dependence intensity (EDI) function as the non-parametric part of the semi-parametric model.

We congratulate our student price winners announced at yesterday's conference dinner 🥳🎊 Sophie Potts won best poster, @Rouven_Michels best presentation and Nikolai Spuck best paper! Very well deserved! 🏆

@Shirinmgh closes this session before everyone can get ready for the conference dinner. She imputes right-censored survival data by using parametric and non-parametric Bayesian imputation. By doing so, one can overcome challenges that are imposed by censored data.

Ever went dolphin watching and wondered how many dolphins you didn't see? 🐬 Pere Puig is presenting work on deriving a lower bound for population size in capture-recapture experiments that yield data that are right-censored by the number of maximum sights.

In the last session of the day, Tathagata Basu conducts Bayesian regression with missing data by using interval imputation. Nonetheless, his method allows to derive point estimates.

Finally, Vito Muggeo presents a segmented quantile regression process with quantile-varying breakpoints. His work was inspired by the association of weight/BMI and the realized maximum speed of animals. 🐁🐅🐘

Michele Lambardi di San Miniato presents Bayesian ordinal regression for crowed-sourced fact-checking. His approach might help us to automatically check if fake news have been told to us! 🔍

Melania Lembo kicks off the afternoon by highlighting the importance of taking context into account. In relational event models, she includes global covariates by defining a time-shifted process.