Assessment and Evaluation of Different Machine Learning Algorithms for Predicting Student Performance.

Area of Science:

• Educational Technology
• Computer Science
• Data Science

Background:

• Student performance is vital for tertiary institution success and university rankings.
• Accurate prediction of student academic achievement faces challenges due to limited research in machine learning (ML) approaches.
• Effective ML tools are needed for modeling and assessing student performance to enhance educational outcomes.

Purpose of the Study:

• To investigate existing ML approaches and key features for predicting student performance.
• To identify the most effective ML models and input variables for student performance prediction.
• To analyze research trends in ML for academic achievement prediction.

Main Methods:

• Systematic literature search of online databases for studies published between 2015 and 2021.
• Evaluation of 39 selected studies on ML applications in student performance prediction.
• Analysis of commonly used ML models and predictive features.

Main Results:

• Six ML models were predominantly used: decision tree (DT), artificial neural networks (ANNs), support vector machine (SVM), K-nearest neighbor (KNN), linear regression (LinR), and Naive Bayes (NB).
• Artificial neural networks (ANNs) demonstrated superior performance and higher accuracy compared to other models.
• Academic, demographic, internal assessment, and family/personal attributes were the most common predictive features.

Conclusions:

• Research in ML for student performance prediction is increasing, with a growing diversity of algorithms applied.
• ML models, particularly ANNs, show significant potential for identifying and improving academic performance.
• The findings suggest ML can be a valuable tool for educators to understand and enhance student success in tertiary education.