SMARTGRAD: Prediksi Kelulusan Tepat Waktu Mahasiswa Kampus Merdeka

Abstract

On-time graduation is a primary indicator of student success and serves as a key benchmark for the quality of higher education institutions. This study aims to develop SmartGrad, a prediction model for on-time graduation based on the Naive Bayes algorithm, supported by feature selection using Decision Tree. The model integrates academic variables (semester GPA, average grades) and non-academic variables (types of MBKM, employment status, age) to produce accurate and contextual predictions. The research dataset comprises 313 entries with 17 attributes, processed through feature selection and classification stages. Evaluation results demonstrate the model's excellent performance, with an average accuracy of 88.8%, precision of 90.5%, recall of 97.9%, and an F1-score of 94.0%. The implementation of SmartGrad as an interactive web application based on Streamlit supports transparent and easily comprehensible decision-making. The novelty of this research lies in the integration of MBKM factors and employment status into the prediction model, as well as the application of an interpretable AI approach to support higher education policies and the achievement of Sustainable Development Goal 4 (Quality Education). These findings are expected to serve as a strategic reference for higher education administrators in enhancing academic quality and the effectiveness of the Freedom of Learning Independent Campus program.