Machine learning techniques for analyzing student’s performance in Islamic Studies

Abstract

Many learning institutions and organizations are currently faced with the acute burden of trying to forecast the academic performance of their students. This paper reflects the application of machine learning tools to discuss the potential and performance of students in Islamic Studies. The framework suggested in this paper, will start with the acquisition of the historical data of the students in the input dataset. First, the forward selection wrapper method is used to select the most meaningful features thus eliminating the redundant qualities in the set of student data. Three types of classifiers are then used to create a classification model based on fuzzy support vector machines (SVM), K-nearest neighbors (K-NN), and Naive Bayes. In such a methodological approach, academic performance is predicted and results measured according to certain criteria. According to results of the experiment, it is noted that feature selection-fuzzy support vector machine (FS-fuzzy SVM) has an excellent accuracy of 99.9% with a sensitivity of 98.50% and a specificity of 98.50% and it is therefore seen to be more effective in predicting the academic performance of students in Islamic Studies.