This project leverages machine learning and natural language processing to predict stock prices. By combining historical market data with sentiment analysis of news headlines, this model provides more accurate and insightful predictions.
- Data Collection & Preparation: Historical stock data and news headlines are collected, cleaned, and preprocessed.
- Sentiment Analysis: News headlines are analyzed using VADER to calculate sentiment scores.
- Technical Indicators: Indicators like SMA, EMA, MACD, RSI, and OBV are calculated to enhance predictive power.
- Machine Learning Model: A Random Forest Regressor is implemented using a rolling window approach.
- Visualization & Evaluation: Actual vs. predicted prices are visualized, and metrics like MAE, RMSE, and R-squared are used for evaluation.
- High Accuracy: Achieved an R-squared value of 0.9996, indicating a highly accurate model.
- Comprehensive Sentiment Analysis: Incorporated sentiment scores (positive, neutral, negative) from news headlines to improve prediction accuracy.
- Feature-Rich Model: Utilized multiple technical indicators to capture market trends and patterns.