Forecasting the Future of Crypto currency: A Machine Learning Approach for Price Prediction

Abstract

virtual forex is said as one of the economic properties extensively recognized as change foreign money. Crypto currency buying and selling has stuck the attention of buyers as crypto currencies may be viewed as extraordinarily profitable investments. correct charge forecasting is essential for optimizing your crypto currency investment returns. for the reason that price prediction is a time-collection challenge, a hybrid deep mastering version has been proposed to are expecting destiny costs of crypto currencies. The hybrid version integrates a one-dimensional convolutional neural community and a Stacked Gated Recurrent Unit (1DCNN-GRU). Given crypto currency charge information through the years, a one- dimensional convolutional neural network encodes the information right into an excessive- degree identification illustration. Stack gate recursion devices then capture lengthy-time period dependencies in expressions. The proposed hybrid version changed into evaluated on 3 one-of-a-kind crypto currency datasets: Bitcoin, Ethereum, and Ripple. Experimental outcomes show that the proposed 1DCNN-GRU version outperforms present techniques with the smallest RMSE values of forty-three.933 for Bitcoin dataset, three.511 for Ethereum dataset, and 0.00128 for Ripple dataset.

Keywords: Machine Learning, Crypto Currency, Forecast, Bitcoin, Arima

JEL Classification: O3