A hybrid AI approach for predicting academic performance in RBE students.

Area of Science:

• Educational Technology
• Artificial Intelligence in Education
• Machine Learning Applications

Background:

• Machine learning (ML) is increasingly used in higher education for data analysis.
• Limited research exists on ML for predicting academic performance in basic education, especially in private denominational institutions.
• This gap highlights an opportunity to develop predictive models for these specific educational settings.

Purpose of the Study:

• To propose a novel predictive approach using machine learning techniques for regular basic education.
• To develop a decision-support tool to aid in educational planning and intervention within private denominational schools.
• To evaluate the effectiveness of various ML and deep learning models for performance prediction.

Main Methods:

• Analysis of multiple machine learning models: Logistic Regression, Support Vector Machine, and Random Forest.
• Evaluation of deep learning models: AlexNet, Gated Recurrent Unit (GRU), and Bidirectional Gated Recurrent Unit (Bi-GRU).
• Development and assessment of ensemble models combining ML and deep learning techniques.

Main Results:

• The Ensemble model, integrating deep learning and machine learning, demonstrated superior performance.
• Key performance metrics including accuracy, precision, and sensitivity were significantly higher in the Ensemble model.
• Comparative analysis showed the Ensemble approach outperforming individual ML and deep learning models.

Conclusions:

• The proposed Ensemble model serves as an effective decision-support tool for predicting student performance in basic education.
• Machine learning and deep learning integration offers a powerful strategy for educational data analysis in specialized institutional contexts.
• Further research can explore the scalability and adaptability of this approach in diverse educational environments.