Early Prediction of Student Learning Performance Through Data Mining: A Systematic Review

Area of Science:

• Educational Data Mining
• Learning Analytics

Background:

• Early prediction of student learning performance is a significant research area.
• Data mining techniques offer valuable tools for this prediction.

Purpose of the Study:

• To provide a comprehensive overview of current research in early prediction of student learning performance.
• To synthesize findings on data mining techniques, variables, and prediction accuracy.

Main Methods:

• A systematic literature review was conducted.
• Papers were identified through major search engines and selected based on predefined criteria.

Main Results:

• 82 selected papers from an initial 133 were analyzed.
• Studies focused on online and face-to-face learning in secondary and tertiary education.
• Commonly used algorithms included J48, Random Forest, SVM, Naive Bayes, logistic, and linear regression.

Conclusions:

• Student assessment and Learning Management System (LMS) interaction data are key predictors.
• The feasibility of early prediction is influenced by the educational system's structure.
• Machine learning algorithms are effective for predicting student outcomes.