Estimation of parameters in nonlinear models with dependent data
Abstract
This dissertation focuses on the analysis of dependent data from repeated measurements study. In the first three chapters, we discuss a method of fitting a nonlinear model to unbalanced or incomplete data from a repeated measurements design. We propose a data transformation method which transforms dependent data into independent data. An EM algorithm is used to estimate unknown within-subject correlation. We also develop a variable selection procedure for adding additional variables into the nonlinear model. An application of this method on a Boys Camp Calcium data set is presented. In Chapter 4, a generalized semi-Markov model is fit to longitudinal survey data. The response variable is assumed to follow the Weibull distribution. We used the coefficient of variation to correct the hypothesis testing conclusions drawn from the dependent data. The application of using the coefficient of variation on a hospitalization data is presented.
Degree
Ph.D.
Advisors
McCabe, Purdue University.
Subject Area
Statistics
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.