SEGMENTASI PENDUDUK MISKIN DI INDONESIA MENGGUNAKAN ALGORITMA K-MEANS

The Covid-19 pandemic has led to an increase in the number of poor people in Indonesia due to government policies aimed at curbing the virus's spread. Grouping poverty levels in Indonesia is crucial for policy-making. This study aims to classify poverty data in Indonesia based on attributes such as GKM, GKNM, IkdK, and IKpK, utilizing the K-Means algorithm for data mining. The results reveal three poverty clusters: low, medium, and high. Clustering the data before, during, and after the pandemic without employing binning techniques did not lead to cluster shifts. However, using binning resulted in cluster shifts in certain provinces in 2020 and 2022. Clustering the data during the pandemic peak, towards normalcy, without binning, showed a shift from high to low poverty levels in Maluku and East Nusa Tenggara provinces. On the other hand, applying binning led to a shift from high to low poverty in East Nusa Tenggara province and from medium to low poverty in Bengkulu province. The Silhouette Coefficient, used as an evaluation metric, ranged from 0.54 to 0.59, indicating that the formed clusters have a good interpretation and are close to 1.