Predicting student academic achievement using stacked ensemble learning with deep neural networks and fuzzy-based...

Area of Science:

• Educational Data Mining
• Artificial Intelligence in Education
• Machine Learning for Student Success

Background:

• Predicting student growth is crucial for personalized education and early intervention.
• Current methods may lack the accuracy needed for effective academic planning.
• Data-driven insights can optimize educational strategies and improve student outcomes.

Purpose of the Study:

• To develop and evaluate a novel hybrid method for predicting student academic performance.
• To enhance the accuracy of student growth prediction for improved academic planning.
• To identify key features influencing student success using a fuzzy logic-based approach.

Main Methods:

• Data preprocessing followed by fuzzy logic-based feature selection using Mutual Information (MI) and Analysis of Variance (ANOVA).
• Feature selection is refined using backward elimination feature selection (BEFS).
• Ensemble modeling using Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Multilayer Perceptron (MLP), with a meta-model for final prediction.

Main Results:

• The proposed methodology achieved significant improvements in predictive accuracy.
• Root Mean Square Error (RMSE) was reported at 0.6%, and Mean Absolute Percentage Error (MAPE) at 0.03%.
• The ensemble model demonstrated superior performance in predicting student academic growth.

Conclusions:

• The novel hybrid approach offers a valuable tool for data-driven academic planning.
• The method provides comprehensive insights into student needs for customized educational programs.
• Accurate prediction of student growth facilitates timely interventions and enhances overall academic performance.