Perbandingan Performa Cluster Model Algoritma K-Means Dalam Mengelompokkan Penerima Bantuan Program Keluarga Harapan

Abstract

Poverty has so far played a role as a problem faced by residents of the Mentawa Baru sub-district, Ketapang. The inability of this community is related to the need to meet education and health needs in social welfare. In assisting the grouping of beneficiary data is carried out using the K-Means algorithm. Apart from that, to increase performance, those who have gone through the first grouping process are then continued using feature selection in the decision tree tool. The algorithm used aims to classify PKH beneficiary data to help the government find out about the handling of the aid program in Mentawa Baru Ketapang sub-district. As for the results obtained from this study, namely, the accuracy of the initial clustering obtained a DBI value of -0.994 at K=8 while the second clustering value that had gone through feature selection with K=3 obtained a DBI value of -0.865. It is known from the performance testing of the comparison of the two clustering that the best performance value is found in the second cluster after going through feature selection.