Application of k-Means Algorithm on Clustering Poor Population Data for Extreme Poverty Elimination

Abstract

Poverty is one of the social problems faced by almost every country in the world. One of the factors causing poverty has not been resolved, namely in an implementation of social assistance policies, the government's survey of the community is still carried out manually so that it is not right on target. So, this research aims to identify the criteria possessed by each group of poor people resulting from data grouping using the K-Means clustering algorithm. By applying the K-Means clustering algorithm to the data of the Targeting for the Acceleration of the Elimination of Extreme Poverty (P3KE) of Sei Litur Tasik Village and modeling the data clustering of the poor population of Sei Litur Tasik Village. The results of testing and evaluating the K-Means Clustering model on the data of the Acceleration of the Elimination of Extreme Poverty (P3KE) are determined to be 2 optimal clusters with an interia value of 0.40 using the Silhouette Score testing method where cluster 1 rich category is 366 families and cluster 2 poor category is 60 families. Modeling of the data clustering system design using the K-Means clustering method was carried out on Google Collaboratory and assisted by supporting literature. The results showed the accuracy of K-Means clustering of 85.92% which means that the accuracy of the analyzed data can be correctly grouped into the appropriate cluster category.