A0292
Title: Estimation and order selection for multivariate exponential power mixture models
Authors: Zhenghui Feng - Harbin Institute of Technology, Shenzhen (China) [presenting]
Xiao Chen - HongKong Baptist University (Hong Kong)
Heng Peng - Hong Kong Baptist Unversity (Hong Kong)
Abstract: Finite mixture models are promising statistical models for investigating the heterogeneity of a population. Using multivariate exponential power mixture models is considered for multivariate non-Gaussian density estimation and approximation. The penalized-likelihood method with a generalized EM algorithm to estimate locations, scale matrices, shape parameters, and mixing probabilities is proposed. Order selection is achieved simultaneously. Properties of the estimated order have been derived. Although it is mainly focused on the unconstrained scale matrix type in multivariate exponential power mixture models, three more parsimonious types of scale matrix have also been considered. Based on simulation and real data analysis, the performance implies the parsimony of the exponential power mixture models and verifies the consistency of order selection.