Individual Difference Assessment Method Based on Cluster Scale Using a Data Reduction Method

Abstract

This paper proposes a new method for assessing individual differences in time series data from multiple subjects, based on the cluster scale using a data reduction method as a measure for quantitatively detecting differences over subjects. This evaluation method is based on a previously proposed method of measuring individual difference among subjects by a common cluster scale over the subjects. However, it is unique in that we can reduce significant amounts of data by extracting data that affect the obtained common clusters by defining a new distance between the target data and the common clusters using the idea of distance based on maximum likelihood estimation using fuzzy covariance. Also, we demonstrate the effectiveness of the proposed method by showing that measuring the data between different subjects on only the selected data on the common cluster scale yields similar results to those obtained when using all data before selection.