Clustrering of BPJS National Health Insurance Participant Using DBSCAN Algorithm

In the current era of Big Data, getting data is no longer a difficult thing because they can access easily it via the internet, which is open access. A large amount of data can cause many problems in the data, such as data that deviates too far from the average (outliers). The method used to handle outlier data is DBSCAN which is density based clustering. The DBSCAN can be applied in various fields, one of which is the social sector, namely the participation of the JKN BPJS Health in West Nusa Tenggara. This study sees the distribution of BPJS Health participation groups, and to detect outliers so that objects with noise are not included in the cluster. The results of the study using the DBSCAN algorithm show that the optimal epsilon value is between 0.37 points by observing the knee of a curve. and MinPts 3, with the highest silhouette value of 0.2763. The highest JKN BPJS participants are in cluster 1 with 5 sub-districts, the second highest cluster is cluster 3 with 5 sub-districts, while the lowest cluster is cluster 2 with 93 sub-districts. The 13 sub-districts are not included in any group because they are noise data.