Dissertation/Thesis Abstract

Optimal designs for segmented polynomial regression models and web-based implementation of optimal design software
by Wu, XiangFeng, Ph.D., State University of New York at Stony Brook, 2007, 81; 3337619
Abstract (Summary)

The segmented polynomial regression model is a special type of polynomial models in that it typically consists of piecewise polynomial sub models. Little research has been done on the optimal designs for segmented polynomial models, particularly for models with heteroscedastic variances. The primary objective of this paper is presenting the up-to-date findings in the optimal design theory for segmented polynomial models and their applications. It is shown that under certain conditions, D-optimal designs for the entire segmented polynomial models are associated with the individual D-optimal designs for the piecewise polynomial sub models. The relation between the overall D-optimal designs and individual D-optimal designs is established subsequently. In addition, we show that in some cases the locally D-optimal designs for a class of segmented polynomial models with nonlinear parameterization are the same as the D-optimal designs for another class of segmented polynomial models with linear parameterization. Secondarily, we present the construction of a unique optimal-design website, on which various web-based optimal design software are incorporated. The ideas and technologies on the implementation are also covered in details in this paper.

Indexing (document details)
School: State University of New York at Stony Brook
School Location: United States -- New York
Source: DAI-B 69/11, Dissertation Abstracts International
Subjects: Computer science
Keywords: Design software, Segmented polynomial regression
Publication Number: 3337619
ISBN: 978-0-549-91415-0
Copyright © 2020 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy