Clustering Analysis of Subsidized Fertilizer Recipients in 2025 Using K-Means++ and Fuzzy C-Means

This study aims to analyze and compare the performance of clustering models in grouping subsidized fertilizer recipient data in 2025 to support the efficiency and accuracy of government distribution policy targets. The recipient data were processed through feature selection, data transformation, data type conversion, and missing value handling. The K-Means++ and Fuzzy C-Means (FCM) clustering methods are applied with the optimal number of clusters (K) set at eight (K=8) based on validity metric analysis. The model evaluation results show that the K-Means++ algorithm produces better cluster quality than FCM. The internal validity metric assessment for K-Means++ on the K=8 cluster shows a Silhouette Score of 0.756, a Davies-Bouldin Index (DBI) of 0.241, and a Calinski-Harabasz index (CHI) of 7901. Meanwhile, FCM reports S=0.737, DBI=0.424, and CHI=4903. This comparison clearly shows that K-Means++ has advantages in terms of clearer cluster separation and stability. The conclusion of this study is that the K-Means++ algorithm is the most effective model and is recommended for use in grouping recipients of subsidized fertilizer assistance in 2025. The results of this grouping (8 clusters) can present an accurate and useful profile for policymakers in designing more effective fertilizer allocation and distribution priority strategies.