Machine learning approach to student performance prediction of online learning

Area of Science:

• Educational Data Mining
• Machine Learning in Education

Background:

• Student performance is vital for educational system improvement.
• Educational data mining leverages data for better learning outcomes.

Purpose of the Study:

• Propose a machine learning method for predicting student performance in online learning.
• Identify and utilize key learning behavioral indicators for accurate predictions.

Main Methods:

• Constructed eleven learning behavioral indicators from online learning processes.
• Filtered indicators based on correlation with student scores, retaining strongly correlated ones as eigenvalues.
• Trained a logistic regression model with Taylor expansion using selected eigenvalue indicators.

Main Results:

• The proposed logistic regression model demonstrated superior prediction ability over comparative models.
• Found a significant dependency between student learning initiative and learning duration.
• Learning duration significantly impacts student performance prediction.

Conclusions:

• The developed machine learning approach effectively predicts student performance in online learning environments.
• Specific learning behaviors, particularly learning duration and initiative, are critical predictors.
• This research contributes to enhancing educational systems through data-driven insights.