Student Performance Prediction Using Classification Models

ABSTRACT

The education industry has gone through major changes amidst the recent COVID-19 pandemic. Facing unforeseen circumstances, educational institutions were forced to shift to an online learning model rather than an offline, classroom-based learning model. The sudden change in the learning model impacted not only students but also the teaching faculty. Even though many resources are available online, simulating a classroom-like study environment is not an easy task. Hence mapping student performance in the new learning model is an essential task. The main goal of our work is to predict the student performance in the online learning model implemented by many colleges and universities amidst the COVID-19 pandemic. Unlike the previous work in this domain, we are purely focusing on an online study system. An online survey was conducted to collect the data from the students who had undergone the aforementioned learning model for at least one semester. The data set for the research includes features that would have an impact on a student’s performance having various attributes. The model strives to predict a student’s performance with good accuracy and help infer where the online learning model can be improved. Several classifiers such as KNN, Gradient boost, Adaboost, Decision tree, SVM, Gaussian NB were used to classify the student data. To validate the performance of these classifiers we have compared them with the latest state-of-the-art works. The Gradient Boost, Xgboost Classifier, and SVM classifiers returned the highest accuracies, in essence, 97.46, 97.45, and 97.45%, respectively. This indicates that the performance of the students is predictable with the given features. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.