This is a user-friendly Streamlit application that integrates AI-based probabilistic forecasting to predict future stock prices. Powered by Hugging Face's Chronos model and Yahoo Finance, it provides interactive visualizations using Plotly, making it a powerful tool for investors and data enthusiasts.
-
Historical Stock Data Visualization:
- Fetches and displays the past year's price and volume data for any stock ticker.
-
AI-Powered Forecasting:
- Uses the Chronos model from Hugging Face to generate probabilistic forecasts.
- Forecast includes median, low, and high price estimates.
-
AI Agent
- uses ChatGPT-4-mini as AI Agent to make a function calling on the AWS Chronos model, which generates probabilistic forecasting.
- Allows the user to write in natural language which forecast to generate, i.g. any company available in yahoo API and any time horizon.
- User does not need to know the ticker, the AI Agent finds the correct ticker of the company and calls the API.
-
Interactive Visualization:
- Historical prices are shown as a line chart.
- Forecast ranges (low, median, high) are displayed as an interactive Plotly chart with shaded regions.
- Python 3.8 or later
- Streamlit installed
- Yahoo Finance API for stock data
- Hugging Face Transformers library for Chronos model
- Plotly for interactive visualization
- OPENAI_API_KEY, get yours from OpenAI Plattform
- Clone the repository:
git clone https://github.com/your-username/stock-price-forecaster.git cd stock_forecast_ai_agent - Install dependencies
pip install -r requirements.txt - Run the App
streamlit run app.py
- Enter a prompt in natural language like : Get stock price forecast for Microsoft for the next 15 days
- Press the Generate Forecast button.
- View:
- Historical stock price data.
- Forecasted prices with a shaded region indicating uncertainty (low to high estimates)
