Perbandingan Model Klasifikasi C4.5, Naïve Bayes, Support Vector Machine dan K-nearest Neighbor untuk Memprediksi Kelayakan Masyarakat dalam Menerima Bantuan PBI APBD

Abstract

This research evaluates the eligibility of the community to receive APBD Contribution Assistance (PBI) using four classification algorithms: C4.5, Naïve Bayes, K-Nearest Neighbor, and Support Vector Machine (SVM). There is a problem of inaccurate distribution of assistance, which prompted the selection of these four methods with specific considerations, C4.5 (Decision Tree) is known for its clarity and interpretability, providing an easy-to-understand understanding of the factors that influence classification decisions, Naïve Bayes was selected for its efficiency and speed in training and testing, suitable for large datasets and can be updated quickly with new data, K-Nearest Neighbor (KNN) is used for decision making based on local patterns in the data, useful if the eligibility decision is local or related to the surrounding environment while Support Vector Machine (SVM): Selected for its ability to handle complex and non-linear datasets. The results show that SVM has the highest Weighted Mean Precision, reaching 91.67%, confirming its superiority as the best choice. These findings make a significant contribution to improving the accuracy of determining the eligibility of PBI APBD beneficiaries, supporting targeting accuracy, and ensuring the effectiveness of the assistance program for people in need.