This repository showcases a robust deep learning model designed to forecast Bitcoin prices using historical data. Leveraging the power of Long Short-Term Memory (LSTM) networks, this project aims to provide accurate predictions that can aid traders, investors, and enthusiasts in making informed decisions.
- Data Acquisition: Utilizes the
yfinancelibrary to fetch historical Bitcoin price data from Yahoo Finance. - Data Visualization: Provides clear visualizations of training and testing datasets to understand price trends.
- Data Preprocessing: Implements Min-Max scaling and sequence creation to prepare data for the LSTM model.
- Deep Learning with PyTorch: Builds and trains a multi-layer LSTM model using PyTorch.
- Model Evaluation: Tracks training and validation loss over epochs to monitor model performance.
- Prediction Visualization: Compares actual vs. predicted Bitcoin prices with comprehensive plots.
- Scalability: Designed to handle large datasets and can be extended to other cryptocurrencies or financial instruments.
The model was trained for 100 epochs, achieving promising results in both training and validation phases. There's a plot in the project that compares the actual Bitcoin prices with the model's predictions, demonstrating the model's accuracy and reliability.
