Stock Predictor UI

This project provides desktop applications for viewing and predicting stock prices using Python and Tkinter. The main application shows actual vs predicted stock prices with a 7-day forecast, while additional utilities are available for stock viewing and advanced LSTM modeling.

How to Run

1. Install dependencies (if not already installed):

   C:\stockPredictorBeta\venv\Scripts\pip.exe install yfinance pandas numpy scikit-learn matplotlib torch

2. Run the main Stock Predictor UI:

   C:\stockPredictorBeta\venv\Scripts\python.exe C:\stockPredictorBeta\stock_predictor_ui.py

3. Usage:

   • The application displays a list of top 10 US stocks for reference
   • Enter a stock ticker (e.g., AAPL, MSFT, etc.)
   • Click "Show Graph" to display actual vs predicted prices with 7-day forecast
   • The graph shows historical data from Jan 2025 - Jun 2025 plus a 7-day forecast

Available Applications

1. Stock Predictor UI (stock_predictor_ui.py) - Main Application

• Purpose: Shows actual vs predicted stock prices with 7-day forecast
• Features:
  • Top 10 US stocks reference list
  • Linear regression prediction model
  • 7-day future price forecast
  • RMSE error calculation displayed in graph title
• Date Range: Jan 2025 - Jun 2025
• Run Command: C:\stockPredictorBeta\venv\Scripts\python.exe C:\stockPredictorBeta\stock_predictor_ui.py

2. Stock Viewer (stock_viewer.py) - Simple Viewer

• Purpose: Basic stock price viewing without predictions
• Features: Graph or table view of historical stock data
• Date Range: 2020-2023
• Run Command: C:\stockPredictorBeta\venv\Scripts\python.exe C:\stockPredictorBeta\stock_viewer.py

3. Advanced LSTM Model (main.py) - Research/Development

• Purpose: Advanced neural network model using PyTorch LSTM
• Features:
  • LSTM neural network for stock prediction
  • GPU support (CUDA if available)
  • Train/test split with RMSE evaluation
  • Detailed error analysis and visualization
• Note: This is a script, not a GUI application
• Run Command: C:\stockPredictorBeta\venv\Scripts\python.exe C:\stockPredictorBeta\main.py

Model Logic & Analysis

Stock Predictor UI Model

• Algorithm: Linear Regression (sklearn.linear_model.LinearRegression)
• Input: Previous day's closing price
• Output: Next day's closing price prediction + 7-day forecast
• Evaluation: RMSE (Root Mean Squared Error)

LSTM Model (main.py)

• Algorithm: Long Short-Term Memory neural network
• Framework: PyTorch with CUDA support
• Sequence Length: 30 days
• Architecture: 2-layer LSTM with 32 hidden units
• Training: 100 epochs with Adam optimizer

Limitations

• Linear regression model is basic and may not capture complex market dynamics
• LSTM model requires significant computational resources and training time
• Stock prediction is inherently uncertain due to market volatility
• Models are for educational/demonstration purposes

File Structure

• stock_predictor_ui.py: Main GUI application with linear regression predictions
• stock_viewer.py: Simple stock data viewer (no predictions)
• main.py: Advanced LSTM model for research/development
• README.md: This documentation file
• venv/: Python virtual environment with dependencies