CREDIT CUSTOMER SEGMENTATION WITH HIERARCHICAL CLUSTERING AT VARIOUS DISTANCES

Cluster analysis is a multivariate technique that aims to group objects based on similar characteristics. The purpose of this study was to see the grouping and comparison of the results of cluster analysis using the ward linkage method with various distances (Euclidean, Manhattan and Mahalanobis) with an assessment of the 5C variable (Character, Capacity, Capital, Collateral, Condition of Economy) on Bank X mortgage customers. In this study, we will compare the results of grouping using the three methods based on the ratio of standard deviations between clusters and within clusters. The sampling technique used nonprobability sampling method with purposive sampling basis. The sample size in this study was 100 customers. The standard deviation ratio between clusters and within clusters shows that the results of the Ward method grouping with Mahalanobis distance are better than the results of grouping using other methods. The results of cluster analysis using the Ward method with Mahalanobis distance are cluster one consisting of 26 customers with compliance as the highest aspect, cluster two consisting of 30 customers with condition as the highest aspect, while cluster three consists of 44 customers with capital as the highest aspect. The originality of this study is to compare the distances in hierarchical cluster analysis, especially with ward linkage in segmentation of credit customer.