NaviQ Air 🌫️

ML-Powered Air Quality Intelligence for Pune

Built by Wild Impala · Samarth Lad, Mehul Mahajan, Atharva Garud - AMD Slingshot 2026

Navigate Your Air - Real CPCB data, ward-level AQI, XGBoost predictions, and health-first decision tools.

NaviQ Air turns 8,010 hours of real CPCB Pune sensor data into actionable health intelligence. It tells you when to go outside, why your ward is polluted, and what city interventions would reduce AQI the most - all powered by a trained XGBoost model with SHAP explainability.

cd naviq-frontend
npm install
npm run dev
# → http://localhost:5173

The frontend works fully offline - all fallbacks are pre-computed from real CPCB data.

cd naviq-backend
pip install -r requirements.txt
uvicorn main:app --reload
# → http://localhost:8000

NaviQ/
├── naviq-backend/
│   ├── main.py                        # FastAPI — all /api/* endpoints
│   ├── requirements.txt
│   ├── 2024_hourly_data.csv           # 8,010 hourly CPCB readings
│   └── aqi_data_pune_2017_to_2024.csv # 2,922 daily AQI records
│
└── naviq-frontend/
    ├── index.html
    ├── vite.config.js
    ├── public/
    │   └── naviq-favicon.svg
    └── src/
        ├── App.jsx                    # Root layout + LIGHT/DARK theme tokens
        ├── main.jsx
        ├── index.css
        ├── components/
        │   ├── Header.jsx             # Ward selector + dark mode toggle
        │   └── Sidebar.jsx            # Navigation + keyboard shortcuts
        ├── data/
        │   ├── api.js                 # API layer with static fallbacks
        │   └── puneData.js            # Ward coordinates + helpers
        └── pages/
            ├── Dashboard.jsx          # City Overview
            ├── MapView.jsx            # Ward Map
            ├── HistoricalPatterns.jsx
            ├── SourceAttribution.jsx  # ML explainability
            ├── Interventions.jsx      # Alerts & Actions
            └── GoOut.jsx              # Health Decision Engine

• XGBoost trained on 2024 CPCB hourly data · SHAP values for explainability
• CO is 3.51× higher on polluted vs clean days - strongest single AQI predictor
• November 2024: 80% of readings were Severe (AQI ≥ 301) - Diwali + crop burning
• Safest daily window: 6 PM · Peak pollution: 10 AM (morning rush + inversion)

Frontend: React 18 · Vite · Recharts · CSS-in-JS
Backend: FastAPI · Pandas · NumPy · Uvicorn
Data: CPCB Pune via Kaggle · 10,932 total readings

NaviQ Air · Wild Impala · AMD Slingshot 2026
Data: CPCB Pune via Kaggle · 2017–2024 · Not a substitute for official CPCB/SAFAR data