The Shape of Disease: Applying geometric data analysis and other methods to time series survey data

Abstract

In this thesis we examine household survey responses to Owner Questions collected in the vaccination zone of the Serengeti Health Initiative and the surrounding region through statistical methods from epidemiology, such as odds ratios and risk ratios, and multiple methods of geometric data analysis including principal component analysis, multiple correspondence analysis, as well as time series adaption of both principal component and multiple factor analyses. We examine whether there is a significant difference between households inside and outside of the vaccination zone and find a significant difference in awareness of animal rabies cases in their respective villages, among other findings. We then examine the role of the responses in each year in the overall variance of the dataset and find that 2015 surveys are the most representative of the survey population across the time series.