Machine Learning Recommender System for an Enhanced Students Course Selection

Abstract

The challenges students face in choosing their course of choice recently have been overwhelming. This necessitated a machine learning-driven model for students' course selection. In this work, a machine learning recommender system that accepts the student's O-level as input, analyses the student's result and then recommends undergraduate courses to the student based on the student's academic performance is proposed. This research classified the course data into four datasets (Agric2, chemistry, Building/PMT, and biology). A total of 4943 instances was obtained, such that each dataset is a class containing courses with the exact requirements. Random forest classifier and decision tree classifier were used to implement each of the datasets. At the evaluation, the decision tree classifier gave an accuracy of 98%, 99.2%, 98.5% and 99.0% on the different datasets, while the random forest classifier gave 98.9%, 98.9%, 99.2% and 98.8%, respectively. This development has resolved the challenge of selecting the best-fit courses, as the model now accurately recommends courses to students based on their previous academic performance. As the model recommends courses that align with students' abilities and goals, it can help reduce the risk of course failure and improve overall academic success and retention rates.