Stress and Anxiety Level Prediction in Students Using Psychological Metrics and Machine Learning Techniques

Student stress is a pressing issue with significant consequences for academic performance and mental health. This study proposes a predictive framework to classify student stress levels (low, medium, high) based on psychological and academic features such as anxiety, self-esteem, depression, and sleep quality. Utilizing a dataset of 1,100 records, we conducted comprehensive preprocessing, exploratory data analysis, and applied advanced statistical techniques including Principal Component Analysis (PCA) and correlation analysis. A Decision Tree classifier was trained and evaluated, achieving a 90% accuracy with balanced F1-scores across all classes. Key insights reveal that anxiety and self-esteem are the most influential predictors. The results demonstrate the effectiveness of machine learning models in enabling early detection of stress, potentially supporting institutional efforts for mental health intervention.