Klasifikasi Kelayakan Penerima Program Keluarga Harapan Menggunakan Algoritma Random Forest di Desa Pancakarya

Abstract

The Family Hope Program (Program Keluarga Harapan/PKH) is one of the Indonesian government's social assistance programs designed to improve the welfare of low-income households. However, determining eligible beneficiaries remains a challenge due to changes in household socio-economic conditions and the limitations of manual verification. Therefore, an accurate data-driven classification approach is required to support the beneficiary selection process. This study aims to analyze the socio-economic characteristics of PKH beneficiaries and non-beneficiaries, develop a beneficiary classification model using the Random Forest algorithm, evaluate the model's performance, and identify socio-economic factors associated with PKH eligibility. This research employed a quantitative approach using a data mining classification technique. The dataset consisted of 470 household records obtained from the Social Welfare Information System (SIKS) of Pancakarya Village, Jember Regency, including 371 eligible and 99 ineligible households. Data preprocessing involved data cleaning, categorical data transformation using Label Encoding, and data partitioning into training and testing sets with an 80:20 ratio. The Random Forest model was constructed using 100 decision trees and evaluated through confusion matrix, accuracy, precision, recall, and F1-score metrics. The experimental results showed that the proposed model achieved an accuracy of 88.30%, correctly classifying 83 out of 94 testing instances. The model demonstrated excellent performance in identifying eligible households, although its ability to detect ineligible households remained relatively limited. Furthermore, socio-economic variables related to occupation, housing conditions, sanitation, flooring, wall and roof materials, cooking fuel, drinking water sources, and electricity sources were found to contribute to the classification process. Overall, the findings indicate that the Random Forest algorithm can effectively support the evaluation and classification of PKH beneficiaries, providing a more objective and data-driven decision support mechanism for social assistance targeting.