NONPARAMETRIC ESTIMATION IN MULTIVARIATE FINITE MIXTURE MODELS
Open Access
Author:
Benaglia, Tatiana A
Graduate Program:
Statistics
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
August 20, 2008
Committee Members:
Bruce G Lindsay, Committee Member Thomas P Hettmansperger, Committee Chair/Co-Chair David Russell Hunter, Committee Member Hoben Thomas, Committee Member
Keywords:
Density Estimation EM Algorithm Nonparametric Mixture Multivariate Mixture
Abstract:
The main goal of this thesis is to provide a complete methodology to analyze finite multivariate mixture models. Our approach is fully nonparametric and it only requires the coordinates to be independent, conditional on the component membership. Since the number of components in the mixture is assumed to be known, we also developed a technique to select the number of components that best fits the model. In addition, we provide some tools to check if the assumption of conditional independence is reasonable. All the methods are evaluated by simulations, and compared with methodology existing in the literature. We also apply our methodology to two real datasets from cognitive psychology.
