K-Modes Analysis with Validation of the DBI in Grouping Provinces in Indonesia based on Indicators of Poor Households

Abstract

One of the most pressing social problems in the world is poverty. One of the efforts to overcome poverty in Indonesia is to group provinces in Indonesia based on indicators of poor households, so that the government can develop poverty reduction strategies according to provincial groups. Cluster analysis is a process of grouping numbers observations based on certain patterns or similarities from these observations. Cluster analysis is divided into two, namely hierarchical and non-hierarchical cluster analysis. Non-hierarchical cluster analysis that can be used is the K-Modes method. K-modes analysis is clustering which is used to group categorical type data. Based on the results of cluster analysis using the K-Modes method, it was found that cluster 1 is a cluster that meets 9 indicators of poor households so that provinces in cluster 1 need to be prioritized in reducing poverty. The Davies Bouildin Index (DBI) is used to see how good the cluster results are used in the K-modes algorithm. By using the DBI, the best number of clusters was obtained as 2 clusters with a DBI value of 1.94063.