Classifying the characteristics of insurance shares: a k-means clustering approach

Abstract

This study aims to apply the k-means clustering method in understanding the characteristics of insurance shares. The eight issuers are divided into three clusters based on price and rate of return. The k-means method's application shows that each cluster has different characteristics, especially for the price variable. Test with panel data regression also discovers different patterns between clusters 2 and 3 in responding to changes in interest rates. The findings of this study indicate that k-means clustering can be used as an initial analysis to understand the characteristics of issuers that investors can use to increase the optimal probability of return.