Abstract: After the boom and bust of cryptocurrencies’ prices in recent years, it has been increasingly regarded as an investment asset. Because of its highly volatile nature, there is a need for good predictions on which to base investment decisions. Although existing studies have leveraged machine learning for more accurate Cryptocurrency price prediction, few have focused on the feasibility of applying different modelling techniques to samples with different data structures and dimensional features. To predict Cryptocurrency price at different frequencies using machine learning techniques, we first download the dataset from a trusted website which keeps all the data of various cryptocurrencies then we classify various Cryptocurrencies by the dataset that is according to the available price. We extract the basic trading features acquired from a cryptocurrency exchange are used for 1 month price prediction. Machine learning algorithms including ARIMA and SVR models for Cryptocurrency’s daily price prediction with high-dimensional features achieve an accuracy of 93% and 94% respectively, outperforming more complicated machine learning algorithms. Compared with benchmark results for daily price prediction, we achieve a better performance, with the highest accuracy of the machine learning algorithm of 97%. Our Hybrid Machine learning model including Support Vector Regression and Autoregressive integrated moving average for One month’s Cryptocurrency price prediction is superior to other Machine learning methods, with accuracy reaching 97%. Our investigation of Cryptocurrency price prediction can be considered a pilot study of the importance of the sample dimension in machine learning techniques.
Keywords: Cryptocurrency, Autoregressive Integrated Moving Average (ARIMA), Support Vector Regression (SVR), Price Prediction, Bitcoin, Transactions, Accuracy, Machine Learning (ML), Deep Learning (DL).
|
DOI:
10.17148/IJARCCE.2022.116121
