Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/166844| Title: | Penerapan Algoritma Self Organizing Maps (SOM) dan K-Medoids dalam Penggerombolan Indikator Kriminalitas di Sumatera Utara |
| Other Titles: | Implementation of Self Organizing Maps (SOM) and K-Medoids Algorithms for Clustering Crime Indicators in North Sumatra |
| Authors: | Sartono, Bagus Sadik, Kusman Putri, Azzahra Adelia |
| Issue Date: | 2025 |
| Publisher: | IPB University |
| Abstract: | Kriminalitas merupakan bentuk tindakan melanggar hukum yang berdampak negatif terhadap masyarakat, baik secara ekonomi maupun psikologis. Provinsi Sumatera Utara merupakan salah satu provinsi dengan jumlah kasus kriminalitas tertinggi. Penelitian ini bertujuan untuk menggerombolkan kabupaten/kota di Provinsi Sumatera Utara tahun 2024 berdasarkan indikator kriminalitas menggunakan algoritma Self-Organizing Maps (SOM) dan K-Medoids, melihat efektivitas kedua metode tersebut, serta mengidentifikasi distribusi kabupaten/kota berdasarkan hasil penggerombolan. Hasil terbaik diperoleh dari metode K-Medoids dengan data yang telah melalui proses winsorizing 5%, yang membagi wilayah menjadi dua gerombol. Gerombol 1 terdiri atas 9 wilayah yang didominasi wilayah perkotaan dengan rata-rata kriminalitas lebih tinggi, sedangkan Gerombol 2 terdiri dari 19 wilayah yang mayoritas merupakan kabupaten dengan tingkat kriminalitas yang lebih rendah. Kemampuan K-Medoids dalam membentuk gerombol yang jelas dan seimbang secara visual mendukung interpretasi spasial yang lebih akurat. Penelitian ini memberikan wawasan penting bagi pembuat kebijakan dan aparat penegak hukum dalam merumuskan strategi pencegahan kejahatan yang lebih terarah di Provinsi Sumatera Utara. Crime is a form of unlawful behavior that negatively impacts society both economically and psychologically. North Sumatra Province is one of the provinces with the highest number of crime cases. This study aims to cluster the regencies/cities in North Sumatra Province in 2024 based on crime indicators using the Self-Organizing Maps (SOM) and K-Medoids algorithms, assess the effectiveness of both methods, and identify the distribution of regions based on the clustering results. The best results were obtained using the K-Medoids method on data that had undergone a 5% winsorizing process, which grouped the regions into two clusters. Cluster 1 consists of 9 regions, predominantly urban areas with a higher average crime rate, while Cluster 2 consists of 19 regions, mostly regencies with a lower crime rate. The ability of K-Medoids to form visually clear and balanced clusters supports more accurate spatial interpretation. This study provides valuable insights for policymakers and law enforcement agencies in formulating more targeted crime prevention strategies in North Sumatra Province. |
| URI: | http://repository.ipb.ac.id/handle/123456789/166844 |
| Appears in Collections: | UT - Statistics and Data Sciences |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| cover_G1401211045_5979cf823df34f7fa49fbf7419edb238.pdf | Cover | 381.48 kB | Adobe PDF | View/Open |
| fulltext_G1401211045_41a8c457dda344d6bf5f917ece263f4a.pdf Restricted Access | Fulltext | 1 MB | Adobe PDF | View/Open |
| lampiran_G1401211045_5142f6fe99854372a7c98279e0cca120.pdf Restricted Access | Lampiran | 368.96 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.