Bitcoin_Prediction

Overview

This project aims to predict the next-day Bitcoin price using deep learning techniques, specifically LSTM (Long Short-Term Memory) models. The project involves data preparation, model training, and evaluation to forecast Bitcoin prices accurately.

Project Structure

• File: Bitcoin_Price_Prediction.ipynb
• Contents:
  • Data preparation using YFinance for Bitcoin market data.
  • LSTM model training with 4 layers for price prediction.
  • Visualization of actual vs. predicted Bitcoin prices.
  • Evaluation of model directional accuracy.

Key Steps

1. Data Preparation:

   • Extracted historical Bitcoin market data.
   • Preprocessed data by scaling and creating look-back windows.

2. LSTM Model Training:

   • Configured LSTM model with 4 layers and Adam optimizer.
   • Fitted the model with 100 epochs and full batch size.
   • Evaluated model performance using mean squared error.

3. Model Evaluation:

   • Calculated directional accuracy of the model.
   • Visualized actual vs. predicted Bitcoin prices.

Results

• LSTM model achieved directional accuracy up to 58.25%.
• Optimal model configuration: 4 layers, 100 epochs, 600-day lookback.
• Model shows potential for trend analysis and price prediction in the cryptocurrency market.

Conclusion

• LSTM model demonstrates promise in predicting Bitcoin prices.
• Continuous optimization of model parameters can enhance predictive capabilities.
• Further refinement and fine-tuning can improve forecasting accuracy for future price movements.