Stock Price Predictor

This project leverages Random Forest models to predict both the direction and magnitude of stock price changes. Using historical data from the Nifty50 index (Indian stock market), the project builds two machine learning models:

1. A Random Forest Classifier to predict whether the stock price will increase or decrease.
2. A Random Forest Regressor to predict by how much the price will increase or decrease.

Project Structure

• stock_predictor.ipynb: This notebook contains the core model training logic. It pulls historical stock data, prepares the dataset, and builds both classifier and regressor models using Random Forest. After training, the models are saved for future predictions.

• trying_the_model.ipynb: This notebook loads the saved models and applies them to new stock data to predict both the direction and the magnitude of price changes.

Installation and Setup

Requirements:

• Python 3.x
• Jupyter Notebook
• yfinance for stock data
• pandas for data manipulation
• scikit-learn for machine learning models
• joblib for saving/loading models

Install the required libraries:

pip install yfinance pandas scikit-learn joblib

Running the Project:

1. Training and testing the Model: stock_predictor.ipynb --> This notebook pulls historical data from the Nifty50, trains a Random Forest Classifier and Regressor, and saves the models as stockpred.pkl and price_change_predictor.pkl.

2. Use the Model: Open and run trying_the_model.ipynb. This notebook loads the saved models and tests them on new stock data. Modify the new_data DataFrame in this notebook to test predictions on your own data.

How the Models Work

• Classifier: Predicts whether the stock price will go up or down based on features such as Close, Volume, Open, High, and Low.

• Regressor: Predicts the magnitude of the price change (increase or decrease) based on the same features.

Example Output:

From trying_the_model.ipynb:

new_data = pd.DataFrame({
    'Close': [25127.95],
    'Volume': [206400],
    'Open': [25023.45],
    'High': [25159.75],
    'Low': [25017.50]
})

output

Predictions for new data: [0]
the stock price will Decrease by [-62.25100365]

Key Features

• Random Forest Classifier: Predicts price movement direction (Increase/Decrease).
• Random Forest Regressor: Predicts how much the price will change.
• Uses Nifty50 stock market data for model training.
• Models are saved and reused in future predictions.

How to Improve

• Tune the hyperparameters of the models to improve accuracy.
• Add more predictors such as moving averages or external factors like market sentiment.
• Implement a more advanced method of feature engineering to reduce RMSE in the regression model.

Conclusion

This project provides a basic but functional model for predicting daily stock price movements and changes using machine learning. It's a great starting point for building more advanced financial models.