Prediksi Kelulusan Siswa Berdasarkan Nilai Akademik Menggunakan Algoritma Perceptron

Abstract

Early monitoring of the risk of student non-graduation at the secondary school level is important so that learning interventions can be carried out in a timely and data-driven manner. This study aims to apply the Perceptron algorithm as a simple machine learning method to predict student graduation status using academic score data, namely UTS and UAS. This approach was chosen because it is lightweight, easy to interpret, and relevant to support educational decision-making. The evaluation of model performance was carried out using accuracy, precision, recall, F1-score, and confusion matrix metrics. The results showed that the Perceptron model was able to classify students who passed with high precision of 0.86, but had limitations in recognizing students who did not pass which was reflected in the recall value of 0.60. An overall accuracy of 0.55 and an F1-score of 0.71 indicate that the model's performance is still affected by data imbalances and limitations of linear separation. These findings suggest that Perceptron has potential as an early tool for predicting graduation, but improved data quality, feature additions, and the use of nonlinear models are recommended for further research.