Perbandingan Algoritma Machine Learning dalam Memprediksi Kelulusan Mahasiswa

Abstract

Penelitian ini bertujuan untuk memprediksi kelulusan mahasiswa Program Studi Informatika Universitas Muhammadiyah Sidoarjo menggunakan algoritma klasifikasi Machine Learning, yaitu Naïve Bayes, Decision Tree, dan Random Forest. Data yang digunakan merupakan data akademik mahasiswa angkatan 2020–2021, mencakup nilai IPS dan jumlah SKS dari semester 1 hingga 6. Proses analisis mengikuti tahapan CRISP-DM, mulai dari pemahaman bisnis hingga evaluasi model. Evaluasi dilakukan menggunakan confusion matrix serta pengukuran akurasi, presisi, recall, dan F1-score untuk membandingkan performa tiap algoritma. Hasil menunjukkan bahwa Random Forest memiliki akurasi tertinggi yaitu 97.50% pada skenario 80:20, disusul Decision Tree dengan 96.25%, dan Naïve Bayes sebesar 86.25%. Selain itu, Random Forest juga mencatatkan nilai presisi dan recall yang tinggi serta F1-score sebesar 97%, menunjukkan kestabilan dan keunggulan model dalam menangani data akademik. Berdasarkan temuan ini, Random Forest dinilai paling optimal dan direkomendasikan untuk digunakan sebagai sistem pendukung keputusan dalam memantau kelulusan mahasiswa secara prediktif dan akurat. This study aims to predict student graduation in the Informatics Study Program at Universitas Muhammadiyah Sidoarjo using Machine Learning classification algorithms, namely Naïve Bayes, Decision Tree, and Random Forest. The dataset consists of academic records from the 2020–2021 cohort, including GPA scores and the number of credits (SKS) taken from semesters 1 to 6. The data analysis process follows the CRISP-DM methodology, covering business understanding, data understanding, data preparation, modeling, evaluation, and deployment. Model evaluation is carried out using confusion matrices along with accuracy, precision, recall, and F1-score to compare the performance of each algorithm. The results show that Random Forest achieved the highest accuracy of 97.50% in the 80:20 scenario, followed by Decision Tree at 96.25%, and Naïve Bayes at 86.25%. In addition, Random Forest demonstrated high precision and recall values with an F1-score of 97%, confirming its stability and effectiveness in academic data classification. Based on these findings, Random Forest is considered the most optimal algorithm and is recommended as a decision support tool for accurately monitoring and predicting student graduation in higher education institutions.