A0172
Title: On regime changes in text data using hidden Markov model of contaminated vMF distribution
Authors: Yingying Zhang - Western Michigan Univesity (United States)
Shuchismita Sarkar - Bowling Green State University (United States) [presenting]
Xuwen Zhu - The University of Alabama (United States)
Yuanyuan Chen - University of Alabama (United States)
Abstract: A novel methodology is presented for analyzing temporal directional data with scatter and heavy tails. A hidden Markov model for contaminated von Mises-Fisher distribution is developed. The model is implemented using a backward elimination algorithm that provides additional flexibility for using it on contaminated as well as non-contaminated data. The method's utility for finding homogeneous time blocks (regimes) is demonstrated in several experimental settings and two real-life text data sets containing presidential addresses and corporate financial statements, respectively.
