Predicting Faculty Emotions and Job Satisfaction from Facial Expressions in Higher Education Using CNN-SVM Models

The emotions of faculty and their job happiness strongly impact the quality of education and the functioning of the institution. Conventional evaluation methods depend on self-reported surveys, which frequently display personal identity and are limited in their ability to accurately reflect real-time emotional conditions. This study presents a hybrid Convolutional Neural Network-Support Vector Machine (CNN-SVM) model for predicting faculty emotions and job satisfaction based on facial expressions. The system employs an innovative CNN architecture for feature extraction and SVM for effective classification. Experimental analysis utilising Kaggle-sourced datasets illustrates the efficiency of the suggested model, attaining 97.6% accuracy, 96.9% precision, 97.2% recall, and an F1-score of 97.0%, outperforming the performance of independent CNN (93.8% accuracy) and SVM (89.2% accuracy). This study improves prediction precision, enabling institutions to more effectively monitor faculty well-being and execute quick actions. Future research will investigate real-time implementation and multi-modal emotional evaluation. The proposed method outperforms current models by integrating deep learning with statistical classification, yielding a more thorough and precise solution for evaluating faculty emotions.