A Machine Learning Model to Predict the Performance of University Students

Abstract

In this era of education and technology, it is undeniable that there is a growing interaction between machine and humans. Student performance is of prime importance as education is the key to success. At the university of Mauritius, the number of students enrolled in a course does not match the number of students graduating as not every student complete their academic cycle of 3 or 4 years. Some extend their course duration as they have to repeat the whole year or several modules, while others exit with a certificate or diploma since they lack the required number of credits to obtain a degree. Unfortunately, the registration of some students with very low average marks are terminated. This research work investigates a machine learning model to predict the performance of university students on a yearly basis. The model will forecast student performance and help take necessary actions before it is too late. The classification technique is used to train the proposed model using an existing student dataset. The training phase generates a training model that can then be used to predict student performance based on parameters such as attendance, marks, study hours, health or average performance. Different algorithms are evaluated and the classification and prediction algorithms which are more accurate are recommended.