Hybrid AI Approach Combining Decision Trees and SVM for Intelligent Tutoring Systems in STEM Education

Intelligent Tutoring Systems (ITS) play a crucial role in STEM education by providing personalized and adaptive learning experiences. However, enhancing student engagement and performance requires systems that can effectively recognize and respond to students' emotional states. This research introduces a pioneering Hybrid AI Approach that seamlessly integrates Decision Trees and Support Vector Machines (SVM) to achieve this goal. Leveraging Kaggle's Student Data Emotion Recognition dataset, the proposed model is designed to understand and respond to students' emotional states during the learning process. To optimize facial expression data processing, Contrast Stretching is applied for data preprocessing, and Recursive Feature Elimination is employed for feature selection. The hybridization of Decision Trees and SVM allows for a balance between interpretability and high-dimensional processing capabilities, contributing to the model's success. Remarkably, the proposed approach outperforms existing methods, including Convolutional Neural Networks (CNN), Deep Neural Networks (DNN), and the combination of Histogram of Oriented Gradients (HOG) and the C4.5 decision tree algorithm (HOG+C4.5)., achieving an outstanding accuracy of 98.7^%• The robust discriminatory power of the model is further confirmed through ROC curve analysis. Implemented in Python, the approach showcases the practicality and versatility of the proposed model within a widely used software environment. This study not only provides a comprehensive methodology and promising results but also sets the stage for further exploration, refinement, and practical implementation of emotion-aware ITS in STEM education. By offering a personalized and intelligent tutoring system tailored to the intricacies of STEM education, the hybrid model's adaptability and efficiency promise to significantly enhance the educational experience for students in these disciplines.