Air Quality Forecasting and Recommendation System

This project provides an end-to-end system for air quality forecasting, model evaluation, and personalized health recommendations, complete with a graphical user interface (GUI) for easy interaction. The system supports city selection, forecasting via LSTM and classical models, and EPA-based health recommendations tailored to user profiles.

---

Features

1. City Selection

Select from multiple U.S. cities such as Los Angeles, New York, Chicago, Houston, Phoenix, Philadelphia, and Boston.

2. Personalized Health Profile

Users may enter:

• Age (optional)
• Respiratory conditions: None, Mild, Moderate, Severe

3. Air Quality Forecasting

• LSTM-based AQI forecasting (default: 3-day)
• Automatic model tuning and evaluation pipeline
• Baseline and statistical models:
  • Naive Forecast
  • Moving Average
  • ARIMA / SARIMA / Prophet
  • LSTM / GRU / Transformer

4. GUI with Multi-Tab Results

The GUI contains:

• Forecast Results: 3-day AQI visualization
• Model Performance: MAE, RMSE, MAPE, R²
• 5-Panel Analysis Dashboard
• Health Recommendations based on AQI + user profile

---

Project Structure

air_quality/
├── README.md               # Main project documentation
├── requirements.txt
│
├── data/
│   └── epa_aqs_data_2025_cleaned.csv
│
├── src/
│   ├── air_quality_gui.py      # GUI
│   ├── data_handler.py         # Data loading, cleaning, API access
│   ├── models.py               # Forecasting models
│   ├── visualizations.py       # Plotting utilities
│   └── recommendations.py      # Health advisory logic
│
└── notebooks/
    ├── 01_data_collection_POC.ipynb
    ├── 02_model_POC.ipynb

---

Installation

pip3 install -r requirements.txt
python -m pip install --upgrade --extra-index-url https://PySimpleGUI.net/install PySimpleGUI

Systems with NVIDIA GPUs will automatically use CUDA for model training and inference when supported.

---

Running the GUI

Ensure your dataset is located at:

data/epa_aqs_data_2025_cleaned.csv

Run the application:

python src/air_quality_gui.py

---

Core Components

1. Data Handling (src/data_handler.py)

Handles:

• EPA API queries
• Real-time AirNow feeds
• Dataset loading (EPA, Kaggle)
• Preprocessing, validation, resampling
• AQI calculation from pollutant measurements

2. Forecasting Models (src/models/)

Includes:

• Baseline models
• ARIMA / SARIMA / Prophet
• LSTM / GRU / Transformer
• Model evaluation and tuning utilities

3. Visualizations (src/visualizations.py)

• Time series forecasting plots
• Model metric charts
• Full 5-panel analysis dashboard

4. Health Recommendations (src/recommendations.py)

AQI thresholds follow EPA standards:

0–50:      Good
51–100:    Moderate
101–150:   Unhealthy for Sensitive Groups
151–200:   Unhealthy
201–300:   Very Unhealthy
300+:      Hazardous

Personalization includes:

• Age-sensitive recommendations
• Respiratory condition severity
• Exposure risk assessment

---