Comparison of Machine Learning Models in Student Result Prediction

Abstract

Prediction of result of students in a particular subject based on their performance in continuous assessment during the semester can be accomplished by various available machine learning models. Every model has its own advantage and limitation due to the algorithm on which they work. Linear regression models have been used very popularly in the area of predictive analytics. Artificial neural networks have also proven their capabilities in prediction. Deep learning techniques are a trend nowadays in data analytics due to their accuracy and performance. This research paper will present a comparison of performance of five popular machine learning models used in predictive analytics—generalized linear model, multilayer perceptron, gradient boost model, Random Forest model, and deep neural network. We have used the result data of students at DIT University, Dehradun, which comprises of schooling marks, continuous assessment marks, and final marks in a subject. On the basis of schooling marks and continuous assessment marks, all these models will predict the final marks of a student in the subject. We will then compare and present the result and performance of these five models.