Professional air quality monitoring and health risk prediction system with explainable AI, policy simulation, and real-time alerts.

This dashboard is fully responsive and works on all devices:

• ✅ Mobile Phones (320px - 640px) - Touch-optimized interface
• ✅ Tablets (641px - 1024px) - Balanced layout
• ✅ Laptops/PCs (1024px+) - Full feature display

• Python 3.10+
• pip (Python package manager)

1. Clone/Download the project:

cd Air_Pollution_Health_Risk_Project

2. Install dependencies:

pip install -r requirements.txt

3. Run the dashboard:

streamlit run streamlit_app.py

4. Open in browser:

• Local: http://localhost:8503
• Network: http://<your-ip>:8503

When opened on phone:

• ✅ Sidebar collapses to hamburger menu
• ✅ Single-column layout automatically
• ✅ Buttons full-width for easy tapping
• ✅ Charts responsive & zoomable
• ✅ Filters optimized for touch
• ✅ Metrics scaled for readability

Access on Phone:

1. Find your laptop/PC IP: ipconfig (Windows) / ifconfig (Mac/Linux)
2. On phone, visit: http://<laptop-ip>:8503
3. Dashboard automatically adjusts to phone screen

When opened on desktop:

• ✅ Full sidebar navigation
• ✅ Multi-column layouts (2-4 columns)
• ✅ All advanced features visible
• ✅ Detailed tables with horizontal scroll
• ✅ Large, interactive charts
• ✅ Side-by-side comparisons

1. Tap Sidebar to navigate pages
2. Scroll down to see all content
3. Tap filters to customize data
4. Swipe charts to explore
5. Long-press for menu options

1. Use sidebar on left (visible if screen wide enough)
2. Tap hamburger (☰) if sidebar hidden
3. Enjoy 2-3 column layouts
4. Export data using download buttons

1. Browse full navigation in sidebar
2. Use advanced filters for detailed analysis
3. Compare side-by-side visualizations
4. Generate comprehensive reports
5. Deep-dive into technical sections

streamlit run streamlit_app.py --server.port 8504

streamlit run streamlit_app.py --logger.level=debug

Edit responsive_css in streamlit_app.py to customize breakpoints.

• Dataset: air_pollution_50000_rows.csv
• Records: 50,000+ air quality observations
• Features: PM2.5, PM10, NO2, SO2, CO, AQI, health metrics
• Coverage: All-India (states & cities)
• Time Period: Multi-year historical data

• Model: Random Forest Classifier (200 trees)
• Accuracy: 72% (test set)
• Training: 70/30 split with 5-fold cross-validation
• Features: Automated encoding & imputation

• 🟢 Green: AQI < 50 (Safe)
• 🟡 Yellow: AQI 50-100 (Moderate)
• 🔴 Red: AQI 100-200 (Poor)
• 🔴 Very Poor: AQI 200-300
• 🟣 Severe: AQI > 300

All pages support:

• 📊 CSV Download - Filtered or full dataset
• 📄 Text Reports - Formatted summaries
• 📉 Chart Export - PNG/JPG with metadata

Download buttons available in:

• Sidebar (global filtered data)
• Each page (specific reports)
• Reports page (comprehensive exports)

• Internet browser (Chrome, Safari, Firefox, Edge)
• 50MB RAM
• 10MB storage

• Python 3.10+
• 4GB RAM
• 500MB disk space
• Windows/Mac/Linux

streamlit run streamlit_app.py --server.port 8504

pip install -r requirements.txt

1. Check laptop IP: ipconfig (Windows)
2. Ensure both on same WiFi
3. Try: http://<ip>:8503 (not localhost)
4. Check firewall settings

• Mobile browsers might cache old version
• Clear browser cache: Settings → Clear browsing data
• Or use incognito/private mode

For issues:

1. Check terminal for error messages
2. Verify all dependencies installed: pip list
3. Restart dashboard: Ctrl+C then run again
4. Check internet connection

• Use WiFi for faster loading
• Close other apps for smoother scrolling
• Avoid opening too many charts simultaneously

• Tap hamburger (☰) to collapse sidebar for more space
• Landscape orientation gives wider views

• Use modern browser (Chrome/Edge recommended)
• Maximize browser window for best experience
• Dual monitors: Dashboard on one, reference on other

Inside dashboard:

• 📖 Explainability Page: Learn how predictions work
• 📊 Model Performance Page: Understand model reliability
• 💡 Executive Summary: High-level insights
• 🚨 Early Warning: Real-time monitoring

Before using:

• Python 3.10+ installed
• pip install -r requirements.txt run
• CSV file present: air_pollution_50000_rows.csv
• Model file present: model.pkl
• Port 8503 available (or change)
• Tested on your device

✨ 14 interactive pages
✨ 72% accurate predictions
✨ Real-time alerts
✨ SHAP explainability
✨ Policy what-if scenarios
✨ Mobile-first responsive design
✨ Cost-benefit analysis
✨ Data quality tracking
✨ Downloadable reports
✨ Dark-mode ready

Happy analyzing! 🎉

Last Updated: December 19, 2025 Version: 0.1.0