Analysis of individual differences in Kansei evaluation data based on cluster analysis.

Abstract

This paper presents a methodology for analyzing individual differences on Kansei evaluation for a set of product samples. This analysis divides subjects into several groups by each subject's Kansei evaluation data , according to what kinds of Kansei are related on what kinds of design elements. The basic idea is to classify the results of cluster analysis on individual subject's ratings. A similarity matrix of subjects is computed by comparing dendrogram of each subject. The member subjects of a group have similar response patterns, and the subjects belong to different groups have different response patterns over all Kansei and design elements. ArboART, neural network based hierarchical clustering is used for individual clustering. The methodology is applied to analyzing evaluation data of milk carton design.