📈 Time Series Forecasting App

A simple app to load time series data, forecast the future using Facebook Prophet, and visualize results with Streamlit.

Run with Docker

docker-compose up -d

Once running, open your browser at http://localhost:8888

📄 Expected CSV Format

The uploaded CSV file must contain at least:

• ds: Date column in YYYY-MM-DD format
• One or more numeric columns to be selected as the target for forecasting

Example:

ds,sales,temperature
2024-01-01,123,15.2
2024-01-02,150,16.8
2024-01-03,170,14.9

You will be able to choose which numeric column (e.g., sales) to forecast.

Forecast Options

• Select the target column for prediction from your dataset
• Choose how many days to forecast (from 7 to 90 days)
• View results as an interactive time series plot

Project Structure

time-series-forecasting/
├── app/
│   ├── main.py        # Streamlit app
│   ├── model.py       # Forecasting logic using Prophet
│   └── utils.py       # CSV loading and preprocessing
├── Dockerfile         # Container build instructions
├── docker-compose.yml # Docker service definition
├── requirements.txt   # Python dependencies
├── .gitignore         # Ignored files and folders
└── README.md          # Project documentation

Features

• Upload your own time series CSV file
• Select a numeric column to forecast
• Adjust forecast length from 7 to 90 days
• Forecast the future using Prophet
• Interactive plot with Streamlit
• Fully containerized with Docker

Technologies

• Python 3.11
• Prophet
• Streamlit
• Docker

License

MIT License. Feel free to use and adapt.

Screenshots