README2.md

---

# 🌍 AQI Forecasting Web App with Map Visualization

This Flask-based web application allows users to input geographical coordinates (latitude and longitude) and receive air quality forecasts (AQI) for the next three days. It combines a deep learning model with real-time data from the Weatherbit API, and visualizes predictions on an interactive map using Folium.

---

## 🚀 Features

- 🌐 **Interactive Map**: View AQI predictions across different locations.
- 📡 **Real-Time Data**: Integrates live air quality data from Weatherbit API.
- 🧠 **Deep Learning Model**: Predicts AQI for the next 3 days based on PM2.5, PM10, NO2, SO2, CO, and current AQI.
- 📊 **Comparative Visualization**: Shows bar plots comparing model predictions and API forecasts.
- 📁 **CSV Logging**: Stores prediction data and actual AQI values from API into CSV files.

---

## 🛠️ Setup Instructions

### 1. Clone the Repository

```bash
git clone https://github.com/your-username/aqi-forecast-app.git
cd aqi-forecast-app
```

### 2. Install Requirements

We recommend using a virtual environment.

```bash
pip install -r requirements.txt
```

### 3. Add Your API Key

Replace the `API_KEY` variable in `app.py` with your [Weatherbit API key](https://www.weatherbit.io/account/create).

```python
API_KEY = "your_api_key_here"
```

### 4. Add Model and Scalers

Ensure the following model and scaler files are in the project directory:

- `FUTURE_AQI_v1.json`
- `FUTURE_AQI_v1.weights.h5`
- `scaler_X_cpcb_4.pkl`
- `scaler_y_cpcb_4.pkl`

> These are required to load the trained model and scale inputs/outputs.

### 5. Run the App

```bash
python app.py
```

Visit [http://127.0.0.1:5000](http://127.0.0.1:5000) in your browser.

---

## 📂 File Structure

```
.
├── app.py                       # Main Flask app
├── FUTURE_AQI_v1.json          # Model architecture
├── FUTURE_AQI_v1.weights.h5    # Trained model weights
├── scaler_X_cpcb_4.pkl         # Scaler for input features
├── scaler_y_cpcb_4.pkl         # Scaler for output predictions
├── aqi_data.csv                # Stores model predictions
├── aqi_data_actual_api.csv     # Stores actual API forecast data
└── templates/
    └── aqi_forecast_with_legend.html  # HTML template
```

---

## 📊 AQI Color Codes

The app uses the following colors for AQI categories:

| AQI Range      | Color        |
|----------------|--------------|
| 0–50           | Green        |
| 51–100         | Light Green  |
| 101–150        | Orange       |
| 151–200        | Red          |
| 201–300        | Purple       |
| 301+           | Gray         |

---

## 📌 Future Improvements

- 📍 Add support for historical AQI trends
- ⏳ Allow user-defined forecasting range
- 📈 Deploy on cloud for public access

---