Employing artificial bee and ant colony optimization in machine learning techniques as a cognitive neuroscience tool

Area of Science:

• Computer Science
• Educational Technology
• Data Science

Background:

• Academic performance prediction for Information Technology (IT) students is vital for identifying influencing factors and areas for improvement.
• Existing prediction systems face challenges with unbalanced data and algorithm tuning, hindering accuracy.
• Effective student academic prediction systems are needed to support educators in forecasting student outcomes.

Purpose of the Study:

• To enhance the accuracy of academic performance prediction for IT students.
• To address challenges of imbalanced data and optimize machine learning algorithms for prediction.
• To identify key features influencing IT student academic success.

Main Methods:

• Applied machine learning algorithms including Decision Tree (DT), k-nearest neighbor, and XGBoost.
• Utilized the Synthetic Minority Oversampling Technique (SMOTE) to handle imbalanced datasets.
• Employed Ant Colony Optimization (ACO) and Artificial Bee Colony optimization for hyperparameter tuning.
• Analyzed feature correlations using the Kendall Tau correlation coefficient.

Main Results:

• The combination of SMOTE and ACO with the Decision Tree (DT) model demonstrated superior performance in academic prediction compared to other tested models.
• Hyperparameter tuning significantly improved the predictive capabilities of the machine learning models.
• Identified key features positively and negatively impacting IT student academic success through correlation analysis.

Conclusions:

• The proposed approach effectively addresses imbalanced data and algorithm tuning issues in IT student academic prediction.
• The optimized Decision Tree model provides a robust and accurate method for forecasting student academic performance.
• Findings offer valuable insights for educational institutions to develop targeted support strategies for students.