Information theoretic methodology for Kansei engineering: Kansei channel and SPRT

Abstract

This chapter provides an application of some notions in information theory to Kansei engineering toward a mathematical methodology for the analysis in Kansei engineering field. First, we propose a new definition of correlation using the mutual information in information theory. Second, we present a relation between Bayes' updating using the notion of binary channel and sequential probability ratio test (SPRT). Third, we illustrate that, for structural change detection of time series data, the SPRT is a very useful method that meets our intuition or Kansei very well. Moreover, we show the effectiveness of SPRT, comparing with Chow test that is well known as a standard method for the change detection.