Enhanced LSTM with ELU for Stock Market Forecasting

Undergraduate thesis implementing enhanced LSTM model with ELU activation for stock market forecasting, as described in the research paper. The full-stack application includes a Python backend for model training/prediction and a Next.js frontend for visualization.

Project Overview

This thesis demonstrates the implementation of LSTM neural networks enhanced with ELU (Exponential Linear Unit) activation for predicting stock market prices. The ELU activation helps mitigate the vanishing gradient problem and improves convergence compared to traditional Tanh architectures.

Key Features

• LSTM model architecture with ELU activation
• Time series data preprocessing and normalization
• Comparative analysis between baseline LSTM and enhanced LSTM-ELU
• Evaluation metrics: MAE, RMSE, R²
• Visualization of predictions vs actual data
• Hyperparameter tuning for optimal performance
• TensorFlow and PyTorch implementations

Results

Based on the research paper, the enhanced LSTM-ELU model achieved:

• 75.64% reduction in MAE (1.90 vs 7.78)
• 75.37% reduction in RMSE (2.49 vs 10.10)
• R² score of 0.9852 (vs 0.7566 for baseline)

Technologies Used

• Language: Python
• ML Frameworks: TensorFlow/Keras, PyTorch
• Data Processing: NumPy, Pandas
• Visualization: Matplotlib, Seaborn
• Development: Jupyter Notebooks