Machine Learning Models for Predicting Academic Performance

Abstract

Based on today's advanced technology, the process of education does not have to be face-to-face; it can also take place online. This makes education more convenient, but there are still many problems that come with it. Machine learning is a very good tool to predict outcomes. The accurate prediction models can help with decision-making and improve data analysis efficiency. However, choosing the appropriate models and comparing the performance between different models remains challenging. This study uses multiple machine learning models to predict the engagement of online course students based on their features. The three machine learning algorithms applied in this study are logistic regression, decision tree, and random forest. The study pre-processed the dataset first, then used the pre- processed dataset to train the models. Finally, it evaluated the performance of the models according to standards such as accuracy and confusion matrices. The study finds that the tree structure-based methods achieve higher accuracy than logistic regression. The finding suggests that instructors can decrease the dropout rate using the academic performance prediction.