Real-Time Student Dropout Risk Prediction Using Machine Learning
Developed a full-stack predictive analytics platform using Flask, SQLAlchemy, and scikit-learn to analyze and visualize university student dropout risks. The system uses a Random Forest Classifier to assess dropout likelihood based on academic, behavioral, mental health, and financial factors.
🔹 Implemented Monte Carlo simulation for synthetic data generation (1000+ students)
🔹 Engineered features such as GPA, attendance, last activity, mental health score, etc.
🔹 Built scheduled retraining with APScheduler, and visualized performance with Matplotlib and Seaborn
🔹 Achieved dynamic user roles (Admin, Student), risk dashboard, and RESTful API endpoints
🔹 Persisted predictions and model metrics to SQLite, enabling detailed student-level risk factor analysis
Tech stack: Python, Flask, SQLite, Pandas, scikit-learn, Matplotlib, HTML/CSS, REST API
