Clustering of Crime-Prone Areas in East Medan Based on Police Data Using K-Means and DBSCAN Algorithms

Gaizka Pasya Dermawan Sinukaban; Al-Khowarizmi Al-Khowarizmi

doi:10.30829/algoritma.v10i1.29498

Clustering of Crime-Prone Areas in East Medan Based on Police Data Using K-Means and DBSCAN Algorithms

Gaizka Pasya Dermawan Sinukaban, Al-Khowarizmi Al-Khowarizmi

Abstract

The Medan Timur sub-district is one of the high-crime areas in Medan City, recording 853 cases out of 1,308 criminal incidents collected by the Medan Timur Police Sector during the 2023–2025 period. The cases consist of motorcycle theft or curanmor (689 cases, 52.7%), aggravated theft or curat (448 cases, 34.3%), and robbery or curas (171 cases, 13.1%), spread across 20 sub-villages with a range of 13 to 162 cases per sub-village. This study clusters crime-prone areas using K-Means and DBSCAN algorithms and compares their performance through the Silhouette Index (SI) and Davies-Bouldin Index (DBI). The features used include total_kriminal, curanmor, curas, curat, and rata_waktu, normalized using Min-Max Normalization. The optimal number of clusters for K-Means was determined through the Elbow method yielding K=3, while DBSCAN parameters were determined through a KNN Distance Plot yielding eps=0.20 and minPts=2. Evaluation results show that K-Means yields SI=0.4105 (weak category) and DBI=1.2599, while DBSCAN yields SI=0.6788 (moderate category) and DBI=0.4986 on 8 non-noise sub-villages. DBSCAN outperforms K-Means on both metrics with an SI difference of 0.2683 and a DBI difference of 0.7613, although K-Means is superior in coverage by clustering all 20 sub-villages. These findings can be utilized by the Medan Timur Police Sector as a basis for determining priority patrol areas and allocating security resources more effectively.

Keywords: Crime; Clustering; K-Means; DBSCAN; Silhouette Index; Davies-Bouldin Index

Full Text:

PDF

References

Azhari, D. A., Maulita, Y., & Ramadani, S. (2024). Pengelompokan Data Kriminal untuk Menentukan Pola Rawan Tindak Kriminal Menggunakan Algoritma K-Means (Studi Kasus: Polsek Hamparan Perak). Jurnal Informatika, 2(5).

Sarastuti, E., Mahdiana, D., & Primanita, R. (2024). Klasterisasi Tindak Kriminalitas di Provinsi Jawa Barat dengan Menggunakan Algoritma K-Medoids. Bit (FTI Universitas Budi Luhur), 21(1), 84–91.

Anfelia, Y., Ula, M., & Retno, S. (2024). Identification of Environmental Security in Relation to the Crime Case Rate in Simeulue Regency Using Density Based Spatial Clustering Method with Noise (DBSCAN). Jurnal Riset Informatika, 2(1), 2–12.

Furqan, M., Aulia, A., & Sriani. (2022). Algoritma K-Means untuk Segmentasi Kematangan Buah Jeruk Berdasarkan Kemiripan Warna. Jurnal Sains Komputer & Informatika (J-SAKTI), 6(1), 199–208.

Tambunan, A. M. A., Khaira, U., & Hutabarat, B. F. (2025). Implementasi Algoritma DBSCAN untuk Klasterisasi Koperasi Berdasarkan Indikator Keuangan. Jurnal Informatika, 07(01), 38–55.

Rahmadayanti, F., Rahayu, R., & Alam, P. (2023). Penerapan Metode Data Mining pada Kasus Kriminalitas. Jurnal Ilmiah, 15(1), 52–61.

Ramadani, A. D., et al. (2025). Klasterisasi Data Produksi Daging Sapi Menggunakan Algoritma K-Means Orange Data Mining. Jurnal Informatika, 5(1), 109–114.

Anfossi, L., & Chicco, D. (2026). An easy guide to the Davies-Bouldin index for unsupervised internal clustering evaluation. Discover Computing, 29(212). https://doi.org/10.1007/s44424-026-00012-1

Chicco, D., Campagner, A., Spagnolo, A., Ciucci, D., & Jurman, G. (2024). The Silhouette coefficient and the Davies-Bouldin index are more informative than Dunn index, Calinski-Harabasz index, Shannon entropy, and Gap statistic for unsupervised clustering internal evaluation. PeerJ Computer Science. https://doi.org/10.7717/peerj-cs.3309

Rousseeuw, P. J. (1987). Silhouettes: A graphical aid to the interpretation and validation of cluster analysis. Journal of Computational and Applied Mathematics, 20, 53–65.

Ros, F., Riad, R., & Guillaume, S. (2023). PDBI: A partitioning Davies-Bouldin index for clustering evaluation. Neurocomputing, 528, 178–199.

Biantara, B., Rohana, T., & Juwita, A. R. (2023). Perbandingan Algoritma K-Means dan DBSCAN untuk Pengelompokan Data Penyebaran Covid-19 Seluruh Kecamatan di Provinsi Jawa Barat. Jurnal Ilmiah, IV(1), 88–94.

Liu, L., Wang, Q., & Chen, J. (2024). Cluster validity indices for automatic clustering: A comprehensive review. ScienceDirect. https://doi.org/10.1016/j.heliyon.2025.e42482

Mochurad, L., Sydor, A., & Ratinskiy, O. (2023). A fast parallelized DBSCAN algorithm based on OpenMP for detection of criminals on streaming services. Frontiers in Big Data, 6, 1292923.

Wijaya, T. A., et al. (2024). Perbandingan Algoritma DBSCAN dan K-Means Clustering untuk Pengelompokan Data Gangguan PT. PLN UID Kalselteng. Jurnal Informatika, Vol. 6.

Davies, D. L., & Bouldin, D. W. (1979). A cluster separation measure. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1(2), 224–227.

DOI: http://dx.doi.org/10.30829/algoritma.v10i1.29498