Clustering Retirement Mutual Fund

Abstract

The special project of this research is to discover knowledge via clustering Retirement Mutual Funds (RMFs) by their qualitative variables, risk assessments, and types of investment, numerical variables, unit trust, net asset values, returns on one-year investment, NAV and associated fees. The raw data is collected manually from the Internet, then analyzed using the Two-step Clustering Analysis for both types of data (TMCM) proposed by Ming-Yi Shih, Jar-Wen Jheng and Lien-Fu Lai which implements an agglomerative data management algorithm. The Co-Occurrence method is used to create random data selection by converting qualitative data to the equivalent numerical form. 

The clustering results illustrate 4 groups of RMFs, each of which has a distinct characteristic that can be intuitively interpreted. The first group (29.92%) comprises of unsecured assets. The second group (22.22%) represents the most of popularity received. The third group (11.11%) contains those that are popular but low-risk. Finally, the fourth group (36.75%) is non-popular and low-risk RMFs as reflected by the market.