時 間：114年3月11日（星期二下午14:10~15:00）
地 點：管理學院新大樓M240
主講人：孫立憲 教授 (中央大學 統計研究所)
講 題：Change Point Detection and Estimation using Copula-Based Markov Chain Models
Abstract
We consider the copula based Markov Chain Models for describing the sequential data with nonlinear feature. In particular, we focus on detection for the structural change for the obtained data. The proposed problem is very important in real application such as finance, industry, and biology.
In order to tackle this issue, the Bayesian online algorithm for the online change point detection and the maximum likelihood estimation (MLE) for the offline problem are provided. Furthermore, for comparison, the varied copulas including the Clayton and the Joe copulas are studied. The performance of the proposed methods is illustrated through the simulation studies and empirical studies.