Penerapan Jaringan Saraf Tiruan Learning Vector Quantization Untuk Pemetaan Wilayah Berpenduduk Miskin di Provinsi Maluku

Abstract

Badan Pusat Statistik (BPS) stated that the number of poor people in Indonesia reached 28.01 million people based on data as of March 2016. This figure is around 10.86 percent of the national population. Province of Maluku as the third poor contributor of all provinces in Indonesia reached 27.74 percent. Note that, there are 8 of total 11 districts/cities in Maluku which are determined as underdeveloped regions (Kementerian PDT, 2015), Maluku Barat Daya (MBD) is one of them. Based on data from BPS, in 2014 the percentage of poor people in district of MBD reached 28.33 percent being the second highest district in Maluku after Maluku Tenggara Barat (MTB). It is quite difficult make the poverty level of MBD lower, due to a large number of villages in MBD have some economic access isolations because of geographical conditions. Various programs and policies in social and health have been done to solve this poverty problem, but still could not overcome this problem yet.

In this paper we have grouped the districts/cities of Maluku based on poverty factors using Learning Vector Quantization (LVQ) method. The results of this research showed that there are 5 poverty clusters in Maluku. Those are: Cluster 1 consists of Maluku Tenggara Barat, Maluku Utara dan Buru; cluster 2 consists of Maluku Tengah; cluster 3 consists of Kep. Aru, Seram Bagian Barat dan Seram Bagian Timur, cluster 4 consists of Maluku Barat Daya dan Buru Selatan; and cluster 5 consists of Ambon and Tual. Each cluster describes the poverty level with respect to its Partition matrix respectively. The results that we obtained also show that cluster 4 has the highest poverty level.