Identifikasi Pola Spasial-Temporal Tindakan Pencurian di Kota X menggunakan Algoritma Dbscan

Abstract

Community safety is a fundamental element of quality of life, yet the increasing phenomenon of criminal acts, especially theft, requires a deep understanding of spatio-temporal patterns for effective prevention. This study aims to visualize the spatial distribution of theft, provide information for authorities in designing more effective crime prevention strategies, and cluster crime types based on location and time. Methods using a quantitative descriptive approach, secondary crime data (Chicago Crimes 2012-2017) consisting of 1,048,575 rows from Kaggle was processed through pre-processing (handling duplicates, missing values, data inconsistencies) and analyzed using the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm for spatio-temporal clustering, with results visualized using Tableau. The research results identified spatial and temporal patterns of theft in City X from 2012-2017. It was found that theft is widespread, with streets, residences, and parking lots being the most vulnerable locations. Temporally, mid-July to August was the most vulnerable period. Spatial clusters (-1: safe; 0: vulnerable; 1, 2, 3: safe) and temporal clusters (night and day: watched; morning: not many occurrences) were successfully identified. This study confirms that Convolutional Neural Network with Inception-v4 is highly effective for oil palm fruit classification and can be applied in plantation industries to improve harvest efficiency and production quality. The use of AI-based methods like Convolutional Neural Network offers fast, accurate, and reliable solutions for agricultural object classification.