Logistic Regression and Random Forest Comparison in Predicting Students’ Qualification Based on Students’ Half-Semester Performance

The main aim of this paper is to utilize machine learning algorithms for predicting and identifying students who are more likely to fail in university examinations at an early stage, based on their performance during the learning process. The objective is to prevent the risk of failing. This paper compares and reports the use of two machine learning algorithms: Logistic Regression (LR) and Random Forest (RF) for classifying students' success and failure, with the major goal being to achieve better prediction accuracy of the classifier. After conducting experiments using the Laboratory Operations Dataset, which includes valuable features like student attendance, subject difficulty level, and mid-score, the results show that Logistic Regression outperforms Random Forest with an accuracy of 70.8%, precision of 40.4%, and recall of 97.0%, whereas Random Forest achieves an accuracy of 74.7%, precision of 59.8%, and recall of 82.7%. Furthermore, the evaluation metric using the "Not Passed" class recall is applied as the primary focus to evaluate the certainty in predicting students who will fail in their learning process.