Classification of the U.S. business cycle by dynamic linear discriminant analysis

Abstract

Linear discriminant analysis (LDA) was well established by Meyer and Weinberg and by Heilemann and Münch as a technique for the analysis of business cycles. The technique, however, ignores the chronological order of the underlying time series data. This paper presents a dynamic version (DLDA) of linear discriminant analysis and a dynamic measure of separation as additional instruments for business cycle analysis.