Bitcoin Price Prediction using LSTM

This project uses a Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) to predict Bitcoin prices based on historical time series data.

Overview

1. Data Preprocessing

   • The dataset is loaded and filtered to include only relevant Bitcoin price data.
   • The datetime_id column is converted to a proper date format.
   • The data is normalized using MinMaxScaler to improve model performance.

2. Creating Time Series Sequences

   • A function is implemented to transform the dataset into sequences of input features (look_back period) and corresponding target labels.

3. Building the LSTM Model

   • A sequential LSTM model is defined with layers optimized for time series forecasting.
   • The model is compiled with an appropriate loss function and optimizer.

4. Training and Evaluating the Model

   • The model is trained on historical data and validated using a test dataset.
   • Predictions are generated and compared against actual values.
   • The error is measured using Root Mean Square Error (RMSE).

5. Visualization of Results

   • The predicted vs. actual Bitcoin prices are plotted for better analysis.
   • The model’s ability to capture price trends is assessed.

This project demonstrates how deep learning techniques can be leveraged for financial time series forecasting. 🚀