A systematic framework for preparing and enhancing structured data sets for time series analysis

Abstract

This thesis proposes a framework to systematically prepare and enhance structured data for time series analysis. It suggests the production of intermediate derived calculations, which aid in the analysis and rationalization of variation over time, to enhance the consistency and the efficiency of data analysis. This thesis was developed with the cooperation of a major international financial firm. The use of their actual historical financial credit risk data sets significantly aided this work by providing genuine feedback, validating specific results, and confirming the usefulness of the method. While illustrated through the use of credit risk data sets, the methodology this thesis presents is designed to be applied easily and transparently to structured data sets used for time series analysis.