108/11/27李國榮 教授 (國立成功大學統計學系)
Variable selection in finite mixture of regression models with an
unknown number of components
Prof. Kuo-Jung Lee
(Department of Accounting and Statistics, National Cheng Kung University)
In this paper we present a Bayesian framework for finite mixture models to deal with model selection and the selection of the number of mixture components simultaneously. For that purpose, we propose a new reversible jump Markov Chain Monte Carlo algorithm and model each component as a sparse regression model. Our approach is robust to outliers by using a prior that induces heavy tails and works well under multicollinearity and with high-dimensional data. Finally, we apply the framework to cross-sectional data investigating early warning indicators. Our results reveal two distinct country groups for which estimated effects of vulnerability indicators vary considerably.
日 期：108年11月27日(星期三) 16:10~17:00