Abstract

The Government of Indonesia, through the Program Indonesia Pintar (PIP), supports the education of children from underprivileged families. However, the selection process for PIP recipients at PKBM Orange Muara Kelingi is still conducted manually, which is time-consuming and prone to errors. This research aims to design a machine learning-based selection system to improve efficiency and accuracy in determining eligible students for assistance. The research utilizes the Naïve Bayes and K-Nearest Neighbor (KNN) algorithms as classification techniques. Student data, including age, parental income, number of dependents, and KIP/KPS ownership status, serve as the main attributes. The system development follows the Knowledge Discovery in Database (KDD) approach, encompassing data collection, labeling, transformation, modeling, and evaluation. The best-performing algorithm is implemented in a simple web page, allowing users to view classification results in real-time. The findings reveal that the KNN algorithm outperforms Naïve Bayes in terms of accuracy, precision, and recall. This research contributes to supporting the digitalization of Indonesia's education system, particularly in the distribution of PIP assistance.