Date of Completion

7-10-2015

Embargo Period

7-8-2015

Keywords

clustering, incomplete data, missing data, model-based clustering

Major Advisor

Dipak K. Dey

Co-Major Advisor

Ofer Harel

Associate Advisor

Haim Bar

Associate Advisor

See above

Field of Study

Statistics

Degree

Doctor of Philosophy

Open Access

Open Access

Abstract

Several important questions have yet to be answered concerning clustering incomplete data. For example, how can disparate solutions from multiply imputed cluster results be resolved? Additionally, can a model-selection criterion be developed which can detect the correct number of LCA classes after multiple imputation has been performed? Finally, as cluster analysis depends on measures of uncertainty, what is the e ect of missing values on such measures? This thesis presents new theorems, methodologies, and applications which demonstrate solutions to these pressing questions.

Download
COinS