DATA MINING K-MEDOIDS DAN K-MEANS UNTUK PENGELOMPOKAN POTENSI PRODUKSI KELAPA SAWIT DI INDONESIA

Abstract

Oil palm is a type of palm plant that has a production period every year. The largest distribution of oil palm is in Indonesia. Indonesia has a plantation area of 17.32 million hectares. In detail, the production yield is 26.57 tons with a garden area of 8.51 million hectares. According to USDA data, in 2022 it will decline due to various factors. For this reason, it is necessary to classify the potential for oilIpalm production and the chances of successful production at each location of oil palm plantations. With this, a research was conducted on making a clustering system to see the potential for oil palm production by using a combination of 2 methods, namely K-Medoids and K-Means. K-Medoids function to enable clusters according to the data variables with the lowest (Cluster 1) 18 regions, medium (Cluster 2) 5 regions, and high (Cluster 3)/(Cemtroid) 2 regions on potential yield Area, production, and palm oil productivity. TheIK-Means algorithm is used to classify the average land area of 514,885.72 Ha, production of 1,931,882,84 tons and productivity of 3,227,08 Kg/Ha, with the distribution of low, medium and high potential. The combination of the two works very well because each has its own role according to the research needs. From the combination of the 2 methods K-Medoids and K-Means, the results of the three clusters show that the yield of palm oil which has low potential is 72%, 20%, high is 2%. This segmentation is caused by the similarity of plantation characteristics based on the similarty of area, production, and productivity. There are 2 provinces that have the highest production potential of palm oil, namely West Kalimantan and Riau.