Comparative Analysis of Supervised Machine Learning Algorithms for Evaluating the Performance Level of Students

E-learning paradigm is the most successful part in the arena of Educational Data Mining (EDM). People prefer to get a lot of learning materials in the e-learning platforms rather than physical classrooms. In all educational institutions, it becomes a part of students' academic activities to enroll in any of their prescribed online courses. Moodle is a popular management system which paves way for students and teachers to take learning in a comfort zone. A number of parameters such as assignment submission, assessment criteria, number of clicks made by the learners on a single day, enrollment for the exams etc. can be achieved easily in the Moodle platform. Student dataset of an educational institution is taken for the proposed work. Courses selected by the learners are categorized as low, medium and high level courses with respect to its complexity based on few required parameters. This work aims at comparing the accuracy obtained by different Machine Learning (ML) algorithms for the given data set. Among the input, 60% data is taken as training data set and 40% is considered as the testing data. Confusion matrix has been generated using machine learning algorithms namely Support Vector Machines (SVM), k-Nearest Neighbors (k-NN), Logistic regression(LR), and Random Forest. Important metrics such as recall, precision, F-measure and support have been computed and the classification report has been generated for each algorithm. Experimental results produce 97.5%accuracy with Random forest algorithm and it is revealed to be high when compared to the other specified ML algorithms. Thereby, these results are used for analyzing the performance level of the students.