Mental Health Prediction

How it works

1. Academic data → Label encoding, SMOTE, StandardScaler → XGBoost → Depression
2. Workplace survey → Label encoding, SMOTE, StandardScaler → Random Forest + GridSearchCV → Treatment
3. Social media text → TF-IDF (10k features) → Linear SVM → Severe distress

Models were trained in Google Colab, serialized with joblib and are served by a Flask API.