K-Means Clustering On Rice Harvest Data For Planting Season Recommendation In Subak Cepaka, Tabanan

Dewi, Ni Made Cahyani
Hidayat, Ahmad Tri

Compiler
Vol 14, No 2 (2025)

The Subak farming system in Tabanan Regency, Bali, is vital as a primary rice granary but faces challenges in determining optimal planting patterns. Planting decisions based only on inherited experience often do not match climate conditions, reducing productivity and increasing crop failure risks. This study implements the K-Means Clustering algorithm on five years of historical rice harvest data (2020–2024) to generate accurate planting season recommendations. Monthly data were analyzed and grouped into three categories: rainy, dry, and transitional seasons. The clustering results were integrated into a mobile application that provides farmers with accessible recommendations through an interactive interface and visualization. The effectiveness of the clustering model was evaluated using the Silhouette Score, which indicated good separation and cohesion among clusters, while efficiency was assessed through processing time and algorithm simplicity, confirming that K-Means performed the task with minimal computational cost. This system enables farmers to make data-driven planting decisions, optimize productivity, and support sustainable food security in Bali.