A production-grade ML platform that identifies at-risk students using 3 machine learning models with Explainable AI — built and deployed entirely independently as a 1st-year student.
Three machine learning models run in parallel — Logistic Regression, Decision Tree, and Random Forest — each predicting student risk level with confidence scores.
Explainable AI features show exactly why a student is flagged as high risk — not just the prediction, but the reasoning behind it.
A complete dashboard with student management, CSV upload/export, dark mode, search, and real-time ML predictions — all in a single React app.
Frontend SPA with Recharts for data visualisation, Tailwind CSS for styling, React Router for navigation.
Python backend with REST API endpoints for students, auth, CSV upload, and ML predictions. Auto-generates Swagger docs.
Neon Cloud serverless Postgres database. Stores users and student records with SQLAlchemy ORM.
3 ML models trained, evaluated, and serialised with joblib. Logistic Regression, Decision Tree, and Random Forest.
Secure token-based authentication with bcrypt password hashing and protected API routes.
Frontend deployed on Vercel with automatic GitHub CI/CD. Zero-config deployment.
FastAPI backend hosted on Render with environment variable management and auto-deploy on push.
Bar charts, pie charts, and distribution graphs for CGPA and risk analytics — all interactive.
Manually add students or bulk import via CSV with name, register number, department, attendance, and CGPA.
All 3 models run instantly on each student — returning a risk prediction and confidence percentage.
View ROC curves, feature importance, and confusion matrices to understand exactly what's driving risk.
Export full reports as CSV and get AI-generated recommendations for each student's next steps.
Fully deployed and working — no signup required to explore the demo.
We use cookies to improve your experience. Privacy Policy.