Analisis Spasial Pada Pemodelan Faktor-Faktor yang Mempengaruhi Kriminalitas di Indonesia Tahun 2020-2022 dengan Geographically Weighted Panel Regression

Abstract

Crime is one of the socio-economic issues that remains unresolved in Indonesia. Although Indonesia is known as a relatively peaceful country, in reality, many citizens still experience criminal acts. This issue requires immediate attention as it directly affects public comfort and safety. This study aims to analyze the effectiveness of the Geographically Weighted Panel Regression (GWPR) method in addressing spatial heterogeneity when modeling the factors influencing crime rates in Indonesia. Additionally, it seeks to identify the best model for each province by comparing the Fixed Effects Model (FEM) and GWPR. The data used cover 34 provinces in Indonesia over the period from 2020 to 2022. The analytical methods applied include panel data regression and GWPR. The results show that the GWPR model using an adaptive Gaussian kernel is the best-performing model, with an R value of 98.61973% and an AIC of -367.0239. The GWPR model produces different equations for each province, accompanied by varying statistically significant variables. Overall, there are nine variables that generally have a significant effect on crime rates: total population, open unemployment rate, percentage of poor population, average years of schooling, gross regional domestic product (GRDP), human development index (HDI), percentage of youth (ages 15–24) not in school, employment, or training, Gini ratio, and the prevalence of food consumption inadequacy.