Stock Price Prediction Using Prophet and LSTM

This project combines the predictive capabilities of Prophet and LSTM (Long Short-Term Memory) models to forecast stock prices. It integrates sequential data processing from LSTM with the robust forecasting features of Prophet, providing a comprehensive method for financial analysis and prediction.

Features

• Data Preprocessing:
  Fetch and prepare historical stock data using yfinance.

• Prophet Model:

  • Forecasts stock prices for the next 365 days.
  • Interactive visualizations using Plotly.

• LSTM Model:

  • Splits data into training and testing sets.
  • Scales data for better model performance.
  • Predicts closing prices and evaluates using Root Mean Squared Error (RMSE).
  • Compares real vs. predicted prices with visualizations.

• Streamlit Integration:

  • User-friendly interface with sliders for interaction.
  • Efficient data caching for smoother performance.

Technologies Used

• Python Libraries:
  • Streamlit
  • Pandas
  • NumPy
  • TensorFlow (Keras)
  • yfinance
  • Prophet
  • Plotly

Installation

1. Clone the Repository:

   git clone https://github.com/susisarvesh/Stock.LSTM.git  
   cd Stock.LSTM  
   
   

2. Set Up a Python Environment (Optional but Recommended):

   python -m venv venv  
   source venv/bin/activate  # On Windows: venv\Scripts\activate  

3. Install Dependencies:

   pip install -r requirements.txt  

4. Run the Application:

   streamlit run app.py  

5. Access the App:
   Open the URL displayed in the terminal (usually http://localhost:8501).

Example Usage

1. Enter the stock ticker symbol (e.g., AAPL for Apple).
2. View the Prophet model's forecast for the next 365 days.
3. Analyze LSTM predictions for closing prices with performance metrics like RMSE.
4. Compare real vs. predicted prices using interactive Plotly charts.

Key Insights

This project highlights how combining Prophet and LSTM models enhances stock price forecasting by leveraging their individual strengths:

• Prophet excels at capturing trends and seasonality.
• LSTM processes sequential data effectively for accurate predictions.

Disclaimer

This project is for educational purposes only. Stock market predictions made using machine learning models should be supplemented with professional advice and thorough analysis.

Contributions

Contributions are welcome! If you find a bug or have a feature request, feel free to open an issue or create a pull request.

License

This project is licensed under the MIT License. See the LICENSE file for details.

---

Keywords: Streamlit, Pandas, NumPy, Keras (TensorFlow), yfinance, Prophet (Facebook), Plotly, LSTM, Slider (UI interaction), Data caching.

## Conference Paper

[Download Conference Paper 370 (Final)](https://ik.imagekit.io/imagespath/Conference%20Paper%20370%20Final.docx?updatedAt=1739293594265)