KLASTERISASI KERAGAMAN JENIS KONSUMSI PANGAN MENGGUNAKAN ALGORITMA K-MEANS DAN IMPLIKASINYA TERHADAP KEBIJAKAN DAERAH

Abstract

Food consumption is a fundamental indicator of food security and nutritional quality, yet in many developing countries, diets remain monotonous and heavily reliant on staple foods such as rice. In Indonesia, this dependency poses systemic risks, particularly under climate variability, production failures, or supply chain disruptions. To provide evidence-based insights into dietary diversity, this study applied K-Means clustering to household food consumption data across 11 districts and cities in Jambi Province, sourced from the Badan Pangan Nasional (National Food Agency). The analysis involved 8 variables representing major food groups—staple carbohydrates, animal-based foods, legumes, fruits and vegetables, oils and fats, and sugar—standardized through z-scores, with cluster validity evaluated using the Davies–Bouldin Index. The results identified three distinct clusters: (1) simple consumption with low diversity, dominated by rural districts with limited access to nutrient-rich foods; (2) balanced and diverse consumption, mainly urban and coastal areas with broader market access; and (3) high-energy but less varied consumption, characterized by rice- and fat-dominated diets in mountainous regions. These findings reveal both geographic and socio-economic disparities in food consumption, underscoring the need for targeted policies. Cluster-specific recommendations include improving access and nutrition education in rural areas, sustaining balanced diets in urban settings, and promoting diversification through local commodities in highland regions. The study concludes that clustering analysis offers a valuable tool for policymakers to design adaptive, evidence-based food security strategies aligned with local contexts.