WasteWise

WasteWise is a fullstack food-waste prediction and analytics platform for restaurants. It combines a React dashboard, NestJS API, MongoDB persistence, and a FastAPI ML microservice serving a Random Forest pipeline.

Stack

• Frontend: React 18, TypeScript, Vite, Tailwind CSS, Recharts, Framer Motion, Zod
• Backend: NestJS, TypeScript, MongoDB/Mongoose, JWT access and refresh tokens, WebSockets
• ML service: FastAPI, scikit-learn, pandas, RandomForestClassifier pipeline
• Infrastructure: Docker Compose with MongoDB, backend, frontend, ML service

Quick Start

1. Copy environment files:

cp backend/.env.example backend/.env
cp frontend/.env.example frontend/.env
cp ml-service/.env.example ml-service/.env

2. Train the model once. The script auto-detects ../food_wastage_data (1).csv from this workspace, or use DATASET_PATH.

cd ml-service
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
python train.py

3. Start services:

docker compose up --build

4. Open http://localhost:5173.

Local Development Without Docker

cd ml-service && uvicorn main:app --reload --port 8000
cd backend && npm install && npm run start:dev
cd frontend && npm install && npm run dev

API Surface

• POST /api/auth/register
• POST /api/auth/login
• POST /api/auth/refresh
• GET /api/auth/me
• POST /api/ml/predict
• GET /api/predictions
• GET /api/predictions/:id
• DELETE /api/predictions/:id
• GET /api/predictions/export/csv
• GET /api/analytics/summary
• GET /api/analytics/waste-trend
• GET /api/analytics/by-event-type
• GET /api/analytics/by-food-type
• GET /api/analytics/top-factors
• GET /api/analytics/seasonality-food-heatmap

Production Notes

• Replace all JWT secrets before deployment.
• Serve the frontend from a static host/CDN and restrict FRONTEND_ORIGIN.
• Store refresh tokens as hashes only; this implementation never persists raw refresh tokens.
• Train and version the ML pipeline before deployment. Keep models/rf_pipeline.pkl out of git.
• Use MongoDB Atlas or a managed MongoDB instance for production.