Implementasi Sistem Data Mining Penerimaan Siswa Baru Menggunakan Metode Naive Bayes

Abstract

New student admission is a pivotal initial phase in sustaining educational operations at MTs NU Islamiyah, Tangerang, Banten. Although the institution enrolls students annually, fluctuations between registered and unregistered applicants present significant operational challenges, particularly in institutional policy planning and the optimal allocation of learning infrastructure. This study aims to build a classification model to predict new student admission status using the Naive Bayes method implemented in Python. Predictions were conducted by analyzing historical student attributes, including gender, family economic status, and residential distance to the school. The analyzed dataset comprised 623 training instances and 82 testing instances. The algorithm calculated posterior probabilities for each category based on historical records. The testing results demonstrated that the model achieved strong classification performance for the admitted class, yielding a precision of 87%, a recall of 100%, and an f1-score of 93% (support: 71). Conversely, the model showed poor performance in predicting the non-admitted class, resulting in 0% precision, recall, and f1-score (support: 11).