🚀 Air Quality Prediction Web App

Predicting Air Quality Index (AQI) using city-level air pollution data with a playful cartoon-style web interface and interactive smoke effects.

---

📊 About the Dataset

• Source: data/city_day.csv
• Contains daily air quality measurements for multiple cities, including pollutants (PM2.5, PM10, NO, NO2, NOx, NH3, CO, SO2, O3, Benzene, Toluene, Xylene) and AQI values.
• Used for regression modeling to predict AQI.

---

🎯 Objectives

• Load and preprocess air quality data
• Perform exploratory data analysis (EDA)
• Handle missing values and outliers
• Engineer features (date extraction, encoding)
• Build and evaluate regression models to predict AQI
• Compare model performance with different scaling techniques
• Deploy a web app for interactive AQI prediction

---

📝 Steps Performed

1. Data Loading & Initial Exploration
   • Loaded dataset, checked for missing values and duplicates
2. Feature Engineering
   • Extracted day, month, year from date
   • Encoded categorical city names
   • Dropped unnecessary columns
3. Missing Value & Outlier Handling
   • Filled missing values with median
   • Detected and removed outliers using z-score and IQR
4. Exploratory Data Analysis
   • Visualized distributions and correlations
5. Model Building & Evaluation
   • Built linear regression models
   • Compared three approaches:
     • Outliers removed, non-scaled
     • MinMax scaled
     • Standard scaled
   • Evaluated using MSE, RMSE, R-squared
6. Web App Development
   • Created a Flask backend serving predictions via /predict API
   • Designed a cartoon-style frontend with pastel colors, comic fonts, and a cartoon street background
   • Added interactive smoke effects that follow the mouse pointer
   • Integrated AJAX for real-time AQI prediction and display in a speech bubble

---

🛠 Tech Stack

• Languages: Python, HTML, CSS, JavaScript
• Libraries: pandas, numpy, matplotlib, seaborn, scikit-learn, scipy, joblib
• Frameworks/Tools: Jupyter Notebook, Flask

---

🌐 Web App Features

• Cartoon-style, playful UI with Google Fonts (Comic Neue)
• Input boxes and buttons styled with pastel colors and rounded edges
• Cartoon street background with buildings, clouds, and trees
• Interactive smoke effect (light green, cartoon fart style) that follows the mouse pointer
• AJAX-based prediction: inputs sent to Flask /predict endpoint, result shown in a cartoon speech bubble
• Responsive and immersive user experience

---

📈 Test Results

Three regression tests were performed:

1. Outliers removed - Non Scaled Model
   • Metrics: MSE, RMSE, R-squared
2. MinMax Scaled Model
   • Metrics: MSE, RMSE, R-squared
3. Standard Scaled Model
   • Metrics: MSE, RMSE, R-squared

(Please refer to the notebook output for actual values)

---

🏁 Conclusion

This project demonstrates a complete workflow for predicting AQI using regression techniques and deploying a fun, interactive web app. Feature engineering, outlier handling, and scaling methods were compared to assess their impact on model performance. The web app provides a playful and engaging way for users to interact with the model and visualize predictions.

---

📂 Project Structure

Air quality prediction/
│
├── data/
│   └── city_day.csv         # Raw air quality data
├── Air_quality_prediction.ipynb  # Main notebook
├── app.py                  # Flask backend
├── templates/
│   └── index.html          # Cartoon-style frontend
├── static/
│   └── style.css           # Cartoon CSS styles
├── requirements.txt        # Python dependencies
├── README.md               # Project documentation
└── LICENSE                 # License file

---

🤝 Contributing

Contributions are welcome! Please open issues or submit pull requests for improvements, bug fixes, or new features.

---

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.

---

👤 Contact / Credits

Author: Devarsh S R

---

✨ Feel free to star the repo and share your feedback! ✨