报告简介  Abstract

Hidden Markov model (HMM) is a useful tool for simultaneously analyzing a longitudinal observation process and its dynamic transition process. Existing HMMs have mainly focused on mean regression for the longitudinal response. However, tails of the response distribution are as important as the centre in many substantive studies. We propose a quantile HMM to provide a systematic method to examine the whole conditional distribution of the response given the hidden state and potential covariates. Instead of considering homogeneous HMMs which assume the probabilities of between-state transitions are independent of subject- and time-specific characteristics, we allow the transition probabilities  to depend on exogenous covariates, thereby yielding non-homogeneous Markov chains and making the proposed model more flexible than its homogeneous counterpart. We develop a Bayesian approach coupled with efficient Markov chain Monte Carlo methods for statistical inference. Simulation studies are conducted to assess the empirical performance of the proposed method. An application of the proposed methodology to a cocaine use study provides new insights into the  prevention of cocaine use.

嘉宾简介  About the Speaker

唐炎林，华东师范大学统计学院研究员，博士生导师。2012年1月于复旦大学统计系获得博士学位，师从朱仲义教授。2012年5月起在同济大学数学系工作，历任讲师、副教授，2019年1月加入华东师范大学统计学院。读博、工作期间曾访问何旭铭教授一年，王会霞教授两年，多次访问香港中文大学宋心远教授。主要研究方向为分位数回归、高维数据统计推断、删失数据，主持国家自然科学基金面上项目、青年项目各一项，上海市浦江人才计划一项，在Biometrika、PNAS、Statistica Sinica、Biometrics等SCI期刊发表论文二十余篇。目前担任国际SCI期刊Journal of Korean Statistical Society的副主编。

讲座海报