This is an interactive time series forecasting application built with Gradio. It allows you to train models and make predictions on various time series datasets through a simple web interface.

✨ Features

Upload any time series dataset in CSV format

Select forecasting models (ARIMA, N-BEATS and Bi-iGRU)

Train interactively with adjustable parameters

Visualize predictions in real time

🚀 How to Run

1 - Create and activate a virtual environment:

python -m venv venv 
source venv/bin/activate 

venv\Scripts\activate # On Windows

2 - Install dependencies:

pip install -r requirements.txt

3 - Launch the app:

python app.py

4 - Open your browser at: http://127.0.0.1:7860

---

To use it in Docker:

Build the image (run in repo root)

docker build -t forecasting-app .

Run container (mapping port 7860 on host to port 7860 in container)

docker run -p 7860:7860 forecasting-app