🌀 Time Series Air Quality Forecasting

A time series forecasting project to predict hourly air pollutant concentrations for the next 48 hours using statistical and machine learning models.

📊 Project Objective

To forecast the concentration of air pollutants using a dataset of 9358 hourly records. This project involves:

• Data preprocessing
• Feature engineering
• Time series modeling (ARIMA, Prophet, etc.)
• Residual and error analysis
• Model comparison using RMSE

---

🧰 Tech Stack & Tools

• Python (Pandas, NumPy, scikit-learn)
• Time Series Models: ARIMA, Facebook Prophet, etc.
• Visualization: Matplotlib, Seaborn, Plotly
• Jupyter Notebook for development
• Report: Comparative analysis in PDF format

---

📁 Project Structure

time-series-air-quality-forecasting/
├── data/
│   └── air_quality.csv                  # Input dataset
├── notebooks/
│   └── EDA_and_modeling.ipynb           # Jupyter notebook with all code
├── models/
│   └── prophet_model.pkl (optional)     # Saved models (if used)
├── results/
│   ├── rmse_comparison.png              # Model comparison visuals
│   └── forecast_plot.png                # Forecast output
├── report/
│   └── air_quality_forecasting_report.pdf
├── requirements.txt                     # Python dependencies
└── README.md

---

📈 Models Used

• ARIMA
• Facebook Prophet
• (Optional: LSTM or XGBoost if included)

Each model was trained and validated using the same train-test split. Performance was evaluated using RMSE.

---

📊 Evaluation Metrics

Model	RMSE (Validation Set)
ARIMA	e.g. 35.2 µg/m³
Prophet	e.g. 28.7 µg/m³

(Actual values should be filled in based on your results)

---

🔍 Key Insights

• Feature engineering (lags, rolling means) significantly improved accuracy.
• Prophet outperformed ARIMA due to its handling of seasonality.
• Residual analysis showed white-noise-like behavior, indicating good model fit.

---

🚀 Getting Started

1. Clone the Repository

git clone https://github.com/yourusername/time-series-air-quality-forecasting.git
cd time-series-air-quality-forecasting

2. Install Dependencies

pip install -r requirements.txt

3. Run the Notebook

Open notebooks/EDA_and_modeling.ipynb and run all cells to see the complete analysis and forecast.

---

📄 License

MIT License

---

👤 Author

Vrishali