Optimasi K-Nearest Neighbor dengan Particle Swarm Optimization pada Klasifikasi Pelanggan Listrik Rumah Tangga Bersubsidi

This research optimizes the K-Nearest Neighbor (KNN) method using Particle Swarm Optimization (PSO) for classifying household electricity subsidy power in Gorontalo Province. Using P3KE data with 98,859 records, this research aims to improve the accuracy of classifying 450 VA and 900 VA power for the electricity subsidy program. The research methodology includes data preprocessing, KNN implementation, parameter optimization using PSO, and model evaluation using a confusion matrix. The research results show an accuracy improvement of 1.3% from 83.53% to 84.83% after optimization. The optimized model showed an increase in precision for the 900 VA class from 0.58 to 0.71, although there was a decrease in recall from 0.32 to 0.25. For the 450 VA class, the model maintained a precision of 0.86 with an increase in recall from 0.95 to 0.98.