Higher Education Students Performance and Feature Analysis Using Classification Models Based on Blended Learning Data

Educational institutes and platforms generate an enormous amount of data. The digitization of the learning process also leads us to a variety of data, including interactive features, academic details, and log activity features. This data should be processed through appropriate methods to have useful insight into it. Classifying students according to features, and predicting their performance according to features is one of the major tasks that can be implemented through Machine Learning. This study aims to examine the features of a student's performance and analyze performance using a blended learning dataset through classification techniques. The dataset used in this study is Higher Education IT stream students of MTech(PG-1) and MCA(PG-2). This study also presents a performance evaluation of classification techniques(MLP(Multilayer Perceptron), LR(Logistic Regression), SVM(Support Vector Machine), and RF(Random Forest)) using listed features. The result shows that CGPA, Quiz participation, and forum posts/attentiveness have a high impact on students’ performance and among these 4 classification models MLP has the highest accuracy in classifying students in appropriate classes.