A Data-driven Deep Learning Approach for Bitcoin Price Forecasting

Published in 2023 24th International Conference on Digital Signal Processing (DSP), Rhodes, Greece, 2023

Bitcoin Deployment

Figure 1: Real Time Price Prediction

Bitcoin as a cryptocurrency has been one of the most important digital coins and the first decentralized digital currency. We propose a shallow Bidirectional-LSTM (Bi-LSTM) model, fed with feature engineered data using our proposed method to forecast bitcoin closing prices in a daily time frame. We compare the performance with that of other forecasting methods, and show that with the help of the proposed feature engineering method, a shallow deep neural network out-performs other popular price forecasting models.

ModelMAE TrainRMSE TrainMAE TestRMSE TestMAPE Test
SVR239.83380.78738.69898.220.04
XG Boost106.39170.261502.741829.440.03
Linear Regression250.08383.18562.19691.230.03
LSTM167.27288.14479.77650.810.04
Bi-LSTM160.76279.93414.51561.760.03

For more details, refer to the GitHub repository.

Recommended citation: P. D. Modi, K. Arshi, P. J. Kunz and A. M. Zoubir, "A Data-driven Deep Learning Approach for Bitcoin Price Forecasting," 2023 24th International Conference on Digital Signal Processing (DSP), Rhodes (Rodos), Greece, 2023, pp. 1-4, doi: 10.1109/DSP58604.2023.10167930.
Download Paper