Functional data analysis tools for the livestock science researcher

Abstract

Livestock tracking technology has the potential to collect large amounts of data on the position of each animal that is tagged. The assimilation and interpretation of such data in a rigorous manner can be both computationally demanding and mathematically challenging. A range of statistical methodologies could be applied to livestock tracking data and it is often difficult for the livestock scientist to determine the most appropriate method to use for a particular application. This paper focuses on the method of functional data analysis, which promises to be a very useful approach for describing vast amounts of spatio-temporal tracking data by using relatively few transform coefficients. The functional data analysis approach also captures the inherent spatiotemporal correlations in the tracking record thereby ensuring that potentially critical information is retained in the analysis. The background and basic concepts of functional data analysis will be introduced followed by the analysis of data that compares sheep with and without parasite burden.