abstract and description
Introduction to Discrete Latent Variable Models
October 5, 2023, at 15:00 - 18:00
October 6, 2023, at 10.00 - 13.00
Seminar held by
Francesco Bartolucci
Full Professor at University of Perugia
Where
Room 2062, U7, Piazza dell’Ateneo Nuovo 1, University of Milano-Bicocca
Department of Statistics and Quantitative Methods
The seminars will be also available on WebEx at the following links:
https://unimib.webex.com/unimib-it/j.php?MTID=m4fcbabbddeb7452c1417798ac5386d89 Number: 2740 731 4878
https://unimib.webex.com/unimib-it/j.php?MTID=mbc3a4f8b75c8c573d8bdee5fab528a22 Number: 2741 173 5498
Abstract
The seminar aims to review the class of latent variable models, focusing on versions of these models based on discrete latent distributions. After a general definition of the models at issue and an illustration of the pros and cons of these models, a description of the main estimation methods will be provided with special emphasis on the EM algorithm and MCMC algorithms with data augmentation. Then, as special cases, I will focus on finite mixture, latent class, and Hidden Markov models. Some empirical examples will illustrate such models.
About the Speaker
Francesco Bartolucci is a Full Professor of Statistics in the Department of Economics of the University of Perugia (IT). His main research interests are on Longitudinal and panel data, Latent variable and mixture models, Models for categorical data, and Optimization and Markov chain Monte Carlo algorithms for parameter estimation. On these topics, he published papers that appeared in prestigious journals of Statistics and Economics. He has acted as Associate Editor or Referee for many statistical journals, and he is currently Editor of “Statistical Modelling: An International Journal”.