This project leverages a Long Short-Term Memory (LSTM) deep learning model to predict Ethereum (ETH) prices using historical market data. It preprocesses time-series data, trains the model, and evaluates its predictive accuracy, providing a basis for actionable insights in the volatile cryptocurrency market.

• Data Preprocessing: Handles missing values, normalizes data, and creates sequences for time-series analysis.
• Deep Learning: Implements an LSTM-based neural network for price prediction.
• Visualization: Provides visual comparisons of predicted vs. actual prices and tracks training/validation loss over epochs.
• Model Saving: Includes functionality to save and load trained models for future predictions.

The project relies on the following technologies:

• Python: Core programming language.
• TensorFlow/Keras: Deep learning framework for building the LSTM model.
• Pandas and NumPy: Data manipulation libraries.
• Matplotlib: Visualization of results and training progress.
• Scikit-Learn: Data normalization with MinMaxScaler.

1. Clone the repository: git clone https://github.com/lleahhhh/eth_price_prediction.git cd eth_price_prediction

2. Install the required packages: pip install -r requirements.txt Ensure you have Python 3.7+ and TensorFlow 2.0+ installed.

3. Add the historical Ethereum price dataset:

• Add ETHUSD_1m_Combined_Index.csv to the project directory.

The dataset consists of historical Ethereum price data with the following columns:

• Open time: Timestamp of the data entry.
• Open, High, Low, Close: Prices during the time interval.
• Volume: Traded volume during the time interval.
• The project focuses on the Close price for prediction.

1. Data Pre-processing:

• Convert timestamps to datetime format and set as the DataFrame index.
• Normalize the Close price using MinMaxScaler.
• Create sequences of 60 historical prices for training.

2. Model Development:

• Build an LSTM model with two LSTM layers and Dense layers for price prediction.
• Train the model using an 80/20 train-test split.

3. Evaluation:

• Evaluate the model’s performance on the test set using loss metrics.
• Visualize predicted vs. actual prices.

4. Model Saving and Loading:

• Save the trained model for future predictions using TensorFlow's .h5 format.
• Load the model to ensure compatibility.

Training vs. Validation Loss:

• Monitored over 5 epochs to prevent overfitting. Price Prediction Accuracy:
• Sample of predictions compared against actual prices for validation.
• Visualization demonstrates alignment between predicted and actual Ethereum prices.

• Run the script: python eth_price_prediction.py
• The model will process the data, train the LSTM model, evaluate its performance, and display results.

Contributions are welcome! Please fork the repository and submit a pull request for any enhancements. Thanks for making it this far!