Recurrent Neural Network model to predict the prices of 4 major crytocurrencies 3 secs in the future.
The RNN learns from the time-series dataset.
The four cryptocurrencies are:
The data I have used are:
- Open
- High
- Low
- Close
- Volume
The Close column measures the final price at the end of each interval.
The Volume column is how much of the asset was traded per each interval, In this project, these are 1 minute intervals. So, at the end of each minute, what was the price of the asset.
The code is written in python 3.6 version.
pip3 install tensorflow=1.10
pip3 install pandas
pip3 install numpy
pip install -U scikit-learn
Number of EPOCHS = 10
Batch size = 64
Future period of prediction = 3 sec
Sequence length = 60
Number of nodes in input layer = 128
Number of nodes in output layer = 2
Model type = Sequential
python3 Cryptocurrency.py
Trained weights and biases are saved in the models directory.