K-Means clustering of crime and socioeconomic factors as a basis for public policy formulation in Indonesia

Abstract

Crime is a multidimensional social phenomenon that is closely associated with regional socioeconomic conditions. This study aims to classify the 34 provinces of Indonesia based on combined crime and socioeconomic indicators using the K-Means clustering method to support evidence-based public policy formulation. The analysis utilised secondary data obtained from the Indonesian Central Bureau of Statistics (BPS), comprising three crime indicators—narcotics-related crimes, theft, and physical violence—and three socioeconomic indicators: the Open Unemployment Rate (OUR), Gross Regional Domestic Product (GRDP), and the Human Development Index (HDI). Prior to clustering, all variables were standardised, and the optimal number of clusters was determined using the Elbow and Silhouette methods, which identified a two-cluster solution (K = 2). The results revealed substantial disparities in crime and socioeconomic characteristics across provinces. Cluster 1, consisting of DKI Jakarta, North Sumatra, West Java, and East Java, exhibited higher levels of crime, economic activity, and unemployment. In contrast, Cluster 0 comprised the remaining provinces and was characterised by lower crime rates and less intensive socioeconomic dynamics. The findings indicate that provinces with higher unemployment and greater economic concentration tend to experience higher crime rates, supporting strain theory and previous empirical studies. This study contributes to the literature by integrating multiple crime and socioeconomic indicators within an unsupervised machine learning framework to identify provincial typologies that can inform differentiated, place-based policies for crime prevention and regional development in Indonesia.