Sistem Data Mining Penentuan Prioritas terhadap Penerima Bantuan Bencana Banjir dengan Metode Naive Bayes dan Klusterisasi K-Means (Studi Kasus: Wilayah Cengkareng 2025)

This research develops a ranking system for flood aid recipients in Jakarta, focusing on Cengkareng, by utilizing K-Means and Naïve Bayes algorithms. Data were obtained from Satu Data Jakarta (2025), comprising 158 records with attributes including region, sub-district, village, average water level, affected RWs, families, individuals, and flood events. The analytical workflow encompasses data cleaning and normalization, risk level clustering using K-Means (three categories: high, medium, low), and predictive classification with Naïve Bayes. Model evaluation at training-testing splits of 70:30, 80:20, and 90:10 reveals that the combined K-Means and Naïve Bayes approach achieves the highest accuracy of 98.18%, significantly outperforming conventional Naïve Bayes which reached only 43.47%. This improvement demonstrates the effectiveness of combining both algorithms for complex data classification. The developed system expedites the prioritization process, facilitates local teams in verifying recipient lists, and enhances the precision of aid distribution and evacuation. Field simulations with community members were conducted to assess the system’s practical implementation and ensure direct access to flood risk information. Future development will focus on integrating external variables such as real-time rainfall data and expanding field testing to other regions.