Using Distance Measure to Perform Optimal Mapping with the K-Medoids Method on Medicinal Plants, Aromatics, and Spices Export

Abstract

About 80% of the world's medicinal plants grow in Indonesia. The Negeri Rempah Foundation also said that Indonesia has more kinds of spices than any other country in Southeast Asia. To figure out the best way to export plants, you need to do a study that groups them by their main destination country. This is called "clustering," and it can be done by doing regional mapping. In the clustering process, measuring deviation/distance or distance space is a key part of figuring out how similar or regular data and items are. K-Medoids is one way to group things together. K-Medoids is an algorithm that groups data based on how far apart they are. Distance Measure is a way to measure the distance between two points. It can help an algorithm sort object into groups based on how similar their variables are. The dataset used comes from the Customs Documents of the Directorate General of Customs and Excise on the official website of the Central Bureau of Statistics for the period of 2012-2021 about the Export of Medicinal, Aromatic, and Spices Plants. This study uses mixed measures (mixed euclideandistance), numerical measures (camberradistance), and bregmandivergences (generalizeddivergence). The mapping results are compared with the validation of the Davies Bouldin Index (DBI). With the help of the rapidminer software, a number of tests were done. The results showed that using mixed measures (mixed euclideandistance) with a value of k=4 gave a DBI value of 0.021. Because it gives a DBI value close to 0, the K-Medoids algorithm with mixed measures (mixedeuclideandistance) is thought to work better than other distance measures.