Application of Mini-Batch K-Means Algorithm for Clustering Divorce Verdict Document Data Of Indramayu District Religious Court

Abstract

Divorce occurs because married couples are no longer able to achieve the main goals of marriage. According to data from the West Java Central Bureau of Statistics, Indramayu Regency recorded the highest number of divorces in West Java during 2021-2023. This condition underscores the need for research to categorize the Plaintiff's or Applicant's arguments, as set out in the divorce decision issued by the Indramayu Regency Religious Court. The textual arguments of the Plaintiff or Petitioner will be pre-processed text and term weighting using Term Frequency-Inverse Document Frequency (TF-IDF). The term weighting results will be clustered using the Mini-Batch K-Means method. Mini-Batch K-Means speeds up computation by using a subset of data per iteration. In addition, the initial centroids are randomly initialized using K-Means++. The evaluation of Mini-Batch K-Means is measured based on the Silhouette coefficient, the number of iterations, and the speed of computation time. The results of this study show that Mini-Batch K-Means with random initialization is the best model, with a Silhouette coefficient of 0.5293, 4 iterations, and a running time of 0.0653 seconds. Based on the visualization results for each cluster, 2 topic groups were identified: quarrel and dispute factors, and work, children, and financial factors.