CLUSTER MODELING FOOD SECURITY BASED ON MACHINE LEARNING APPROACH FOR REGIONAL PLANNING IN LAMPUNG PROVINCE

Abstract

. Food security is crucial in ensuring a region's welfare and socio-economic stability. However, economic inequality and poverty often hinder achieving sustainable food security, particularly in regions such as Lampung Province. This research aims to analyze the relationship between Gross Regional Domestic Product at Constant Prices (GRDP ADHK), the Human Development Index (HDI), and the percentage of household expenditure on food as indicators of food security. Using the Random Forest Classifier machine learning algorithm, a quantitative approach is applied to evaluate each indicator's relative influence, and the K-Means Clustering method is used to group regions based on their socio-economic characteristics. The model evaluation results indicate high performance, with an R² value of 0.994 and very low prediction error. The feature importance analysis reveals that food expenditure is the most dominant indicator (45.6%), followed by GRDP ADHK (35.1%) and HDI (19.3%). Decision tree visualization and cluster analysis identify three regional typologies: Cluster 0 (food-vulnerable, middle-income, high consumption), Cluster 1 (high food security, high economic status, and HDI), and Cluster 2 (low economic status, high HDI, efficient consumption). These findings indicate that food security is influenced not only by income but also by social policies and resource distribution. This study provides a strong data-driven foundation for formulating food and social development policies in Lampung Province and recommends targeted interventions based on the identified regional typologies.