A hybrid ML framework combining sentiment analysis and deep learning models to predict stock prices. Using Twitter for sentiment data, LSTM models for time-series forecasting, and GANs for enhanced prediction accuracy.
- Sentiment Analysis: Processes Twitter data using NLTK's VADER to extract sentiment scores (positive, neutral, negative, and compound).
- Technical Indicators: Computes moving averages, Bollinger Bands, MACD, and more for financial analysis.
- LSTM Model: Predicts stock prices using historical data and sentiment features.
- GAN Integration: Enhances predictions by generating synthetic data to improve model robustness.
- Performance Evaluation: Metrics include RMSE, MAE, MAPE, and R² Score.
- Programming Language: Python
- Deep Learning: TensorFlow, Keras
- Sentiment Analysis: NLTK VADER
- Data Visualization: Matplotlib, Seaborn
- Data Handling: Pandas, NumPy, Scikit-learn
-
Achieved RMSE: 0.0926 on test data with LSTM-GAN integration.
-
Visualized actual vs. predicted stock prices for comprehensive evaluation.