Data Mining Using Clustering Algorithm as Tool for Poverty Analysis

Abstract

Poverty in one place can be reduced or minimized if proper poverty alleviation programs are given to the community. In this study different clustering algorithms were evaluated using silhouette index to get the best clustering algorithm to group the households and analyze the poverty data. The k-means algorithm where k=3 outperformed DBSCAN and k-medoids with a silhouette of 0.308. The algorithm produced three groups or clusters and labelled as non-poor, near poor and poor. The result can help policy-makers formulate and implement poverty reduction policies and programs that are clear, reasonable, realistic, and enforceable.