Document Type

Dissertation - Open Access

Award Date

2017

Degree Name

Doctor of Philosophy (PhD)

Department / School

Mathematics and Statistics

First Advisor

Gary D. Hatfield

Keywords

diabetes, GWR, INLA, Lyme disease, MGWR

Abstract

This dissertation focuses on modeling approach for spatial and spatiotemporal data with epidemiological applications. Chapter one gives the general overview of spatial and spatiotemporal data and challenges in the statistical analysis of spatial and spatiotemporal data, and motivation and objectives of the study. Chapter two describes the regression models commonly used in spatial data analysis. Various types of regression methods such as OLS, GWR and MGWR were used to study the association between diabetes prevalence and socioeconomic and lifestyle factors on county level data of Midwestern United States. A new analysis workflow is purposed for regression analysis of spatial data. Chapter three describes recently developed INLA as an alternative of traditionally used MCMC in Bayesian hierarchical models. INLA method was used to identify the best regression model for the spatiotemporal regression analysis of Lyme disease count data with climatic covariates in county-level data in Minnesota. Chapter four gives the contribution of this dissertation and discusses the direction for the future research.

Library of Congress Subject Headings

Spatial analysis (Statistics)
Regression analysis.
Medical geography.
Diabetes -- Middle West.
Lyme disease -- Minnesota.

Description

Includes bibliographical references

Format

application/pdf

Number of Pages

114

Publisher

South Dakota State University

Download

Share

COinS
 

Rights Statement

In Copyright