Predicting Licensure Examination Performance Using Data Mining Techniques

Data mining is the process of identifying anomalies, trends, and correlations in massive historical data sets in order to predict future events. The aim of this paper is to use data mining techniques to compare various classification algorithms such as J48, Random Forest, Naïve Bayes, and Logistic Regression to develop a model that can reliably predict a student's performance on the Licensure Examination for Teachers (LET). This model will be used by the institution as a decision-making tool to increase the percentage of students who will pass the board examination. The datasets were extracted from student records between the years of 2017 and 2019 from the registrar's office, Bachelor in Business Teacher Education (BBTE) program, and Professional Regulation Commission (PRC) website using various attributes namely; students' General Weighted Average (GWA), General and Professional Education subjects, honors they had received, their scholarship information, records from review centers, and LET results. WEKA, a data mining tool was used to analyze the dataset. According to the study's findings, J48 had the highest accuracy percentage, compared to the other three classification algorithms. Therefore, J48 provides the most accurate results in predicting the student board examination performance. The results were validated using a number of metrics, such as precision, recall, f-measure, and Cohen's kappa.