Tampubolon, Wenny Maria (2026) Optimasi Parameter Kohonen Self Organizing Map Dengan Particle Swarm Optimization Untuk Klasterisasi Provinsi Di Indonesia Berdasarkan Data Kriminalitas. Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Cover)
21083010041-cover.pdf Download (1MB) |
|
Text (Bab 1)
21083010041-bab1.pdf Download (90kB) |
|
Text (Bab 2)
21083010041-bab2.pdf Restricted to Repository staff only until 8 March 2029. Download (199kB) |
|
Text (Bab 3)
21083010041-bab3.pdf Restricted to Repository staff only until 8 March 2029. Download (317kB) |
|
Text (Bab 4)
21083010041-bab4.pdf Restricted to Repository staff only until 8 March 2029. Download (1MB) |
|
Text (Bab 5)
21083010041-bab5.pdf Download (74kB) |
|
Text (Daftar Pustaka)
21083010041-daftarpustaka.pdf Download (99kB) |
|
Text (Lampiran)
21083010041-lampiran.pdf Restricted to Repository staff only until 10 March 2029. Download (36kB) |
Abstract
Crime refers to behavior that violates legal rules and social norms and may cause physical, economic, and psychological harm. Data from the Central Bureau of Statistics in 2024 indicate a significant increase in the number of criminal cases in Indonesia, rising from 327,965 cases in 2022 to 584,991 cases in 2023, while only 299,517 cases were successfully resolved. This considerable gap reflects weaknesses in crime handling efforts, highlighting the need for an analysis of crime patterns to obtain a clearer understanding of conditions across provinces. This study aims to cluster provinces in Indonesia based on crime data using the Kohonen Self Organizing Map (SOM) method. SOM is capable of performing unsupervised data clustering by mapping high dimensional data into a two dimensional representation, allowing inter provincial crime patterns to be observed more clearly. To improve SOM performance, Particle Swarm Optimization (PSO) is employed as an optimization method that mimics the social behavior of a group of particles to identify the best parameter combinations. The optimization results show that the optimal SOM parameters are a 3×1 grid size, a sigma value of 0.9, and a learning rate of 0.4. Using these parameters, three clusters are formed with a Davies Bouldin Index value of 0.36, indicating good clustering quality. The distribution of data across clusters consists of cluster 0 with 2 data point, cluster 1 with 31 data point, and cluster 2 with 1 data points. Cluster 1, as the cluster with the largest amount of data, shows relatively even crime characteristics across almost all variables, such as physical crimes, indecency, property rights, narcotics, fraud, and the number of crimes solved, without any one type of crime dominating, thus reflecting the general crime pattern in most provinces.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | Q Science > QA Mathematics Q Science > QA Mathematics > QA76.6 Computer Programming |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Data Science | ||||||||||||
| Depositing User: | Wenny Maria Tampubolon | ||||||||||||
| Date Deposited: | 10 Mar 2026 01:37 | ||||||||||||
| Last Modified: | 10 Mar 2026 02:27 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/50286 |
Actions (login required)
 |
View Item |