Mapping Regional Clusters Based on Total Rescued Food in Indonesia Using K-Means Clustering

Authors

  • Muhammad Rafi Haidar Arsyad Rafi Universitas Sains Teknologi Ekonomi Digital Indonesia
  • Mohammad Riza Radyanto Universitas Sains Teknologi Ekonomi Digital Indonesia
  • Wafi Arifin Universitas Sains Teknologi Ekonomi Digital Indonesia

DOI:

https://doi.org/10.47686/zzrtrt54

Abstract

This study maps regional clusters in Indonesia based on the total amount of rescued food using the K-Means algorithm. The data were obtained from the official dataset Jumlah Total Pangan yang Terselamatkan Update Bulan Januari 2026 published by the National Food Agency through data.go.id. The research followed an IMRAD-oriented workflow comprising data cleaning, data parsing, construction of regional aggregate features, determination of the optimal number of clusters using the elbow and silhouette methods, K-Means clustering, and visualization of cluster results. The analysis identified two optimal clusters, supported by a silhouette score of 0.809828 and a Davies-Bouldin Index of 0.113459. The findings reveal a clear disparity between regions with low and high rescued-food achievements, with West Java emerging as the most dominant region. This study contributes by utilizing an official rescued-food dataset and regional aggregate features as an analytical basis for monitoring, evaluating, and strengthening data-driven food rescue programs in Indonesia.

Keywords : K-Means, Clustering, Rescued Food, Machine Learning, Aggregate Features

Downloads

Published

2026-06-09