Time Series Forcasting And Comparison Using Neural Networks: A Study on Future Cryptocurrency values And Real Currencies

Abstract

Cryptocurrency is a type of currency available only in digital form, not in physical (such as banknote and coins). There are many uses of cryptocurrencies. The most well-known benefit is their ability to send and receive payments at a low cost and at a high speed. Cryptocurrencies, such as bitcoin, act as a censorship-resistant alternative store of wealth that only the individual with the privet keys to the wallet has access to. Hence, no personal bitcoin wallet can ever be frozen by the authorities. Likewise, it is now possible to travel the world by spending cryptocurrencies. As this is an international business all cryptocurrency transactions are done by using US dollars and the currency market is a typical area that presents time-series data and many researchers study on it and proposed various models. Many time-series prediction algorithms have shown their electiveness in practice. The most common algorithms now are based on Recurrent Neural Networks (RNN), as well as its special type Long-Short-Term- Memory (LSTM) and Gated Recurrent Unit (GRU). The aim of this study is to forecast and compare cryptocurrencies such as Bitcoin, Ethereum, Litecoin and Ripple and also real currencies such as Australian dollars, Euro all against US dollars using recurrent neural network.