Online Inference for Dynamic Categorical Regression

The multinomial probit is a widely used model for categorical data in many applications, especially in econometrics and the discrete-choice literature. The computational challenges encountered when fitting this model still motivate ongoing research from the frequentist and Bayesian viewpoints. We consider a dynamic formulation based on a state-space model, where at each time one observes a sample from a multinomial probit whose parameter evolves according to Markovian dynamics, thereby inducing dependence across time. We show that the filtering and predictive distributions of the latent parameter belong to the unified skew-normal family, developing an associated i.i.d.\ sampler to approximate quantities of interest via Monte Carlo. Motivated by the computational bottlenecks of the sampler encountered for larger sample sizes, we also develop approximate methods for online inference based on assumed density filtering and expectation propagation. This gives more scalable, yet accurate, algorithms for online inference about the latent state and prediction of future observations. Finally, we derive a novel expectation maximization procedure to learn from the data the covariance matrix underlying the latent-utility representation of the multinomial probit. We investigate the performance of the proposed methods on simulated data and on a real dataset of reservations from an online booking platform.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-17
DOI
https://doi.org/10.48550/arxiv.2611.22533
Primary Topic
Markov Chains and Monte Carlo Methods
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Online Inference for Dynamic Categorical Regression

Lorenzo Rimella, Augusto Fasano, Giovanni Rebaudo
arXiv (Cornell University)
Markov Chains and Monte Carlo Methods
preprint

Online Inference for Dynamic Categorical Regression

Lorenzo Rimella, Augusto Fasano, Giovanni Rebaudo
preprint en

Abstract

The multinomial probit is a widely used model for categorical data in many applications, especially in econometrics and the discrete-choice literature. The computational challenges encountered when fitting this model still motivate ongoing research from the frequentist and Bayesian viewpoints. We consider a dynamic formulation based on a state-space model, where at each time one observes a sample from a multinomial probit whose parameter evolves according to Markovian dynamics, thereby inducing dependence across time. We show that the filtering and predictive distributions of the latent parameter belong to the unified skew-normal family, developing an associated i.i.d.\ sampler to approximate quantities of interest via Monte Carlo. Motivated by the computational bottlenecks of the sampler encountered for larger sample sizes, we also develop approximate methods for online inference based on assumed density filtering and expectation propagation. This gives more scalable, yet accurate, algorithms for online inference about the latent state and prediction of future observations. Finally, we derive a novel expectation maximization procedure to learn from the data the covariance matrix underlying the latent-utility representation of the multinomial probit. We investigate the performance of the proposed methods on simulated data and on a real dataset of reservations from an online booking platform.

arXiv (Cornell University)
Markov Chains and Monte Carlo Methods
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon