Comparative analysis of artificial neural networks and deep learning algorithms for crypto price forecast

With today's developing technology, there has been an increase in transactions made over the internet. As a result of this, an increase in data has been observed. With this increase in data, many companies are in search of technology for the safe storage, sharing, control and management of data. One of the current technologies is the blockchain structure. The blockchain structure is a distributed database and provides encrypted business tracking consisting of blocks. It is more transparent, no transaction can be changed, high security structure, not centralized structure are the reasons for preference. The blockchain structure is a technology that can be used in many areas, and the most popular usage area today is on cryptocurrencies. With the changing technologies, cryptocurrencies are important investment tools and have had a significant volume in financial markets. It seems that the number of transactions in cryptocurrencies is increasing day by day, and the increase in crypto money prices and the rapid rise in sub-crypto currencies have increased the demand for sub-crypto currencies. In this study, it is aimed to make predictions on Polkadot crypto currency, which is one of the important sub-cryptocurrencies. In the study, the data between 20.08.2020 and 27.02.2023 were used, and according to these data, it was aimed to estimate the daily average Polkadot values as output values. The clusters created for the input values were created in two different ways. In the first input values; Polkadot YouTube search number, Polkadot Google number and Polkadot volume are used. In the second input values, unlike the first input values, Ethereum, the leader of the alt cryptocurrencies, was added. Thus, the effect of Polkadot and Ethereum cryptocurrencies, whose founder is the same person, on each other, more precisely, the effect of Ethereum currency on Polkadot currency has been provided. In this study, which consists of two different input structures, in order to estimate the daily average values of the Polkadot currency, an estimation study was carried out using multi-layered sensors in artificial neural networks and a long-short-term memory structure, which is one of the deep learning methods. When the results are examined, it has been observed that the values consisting of 4 input sets in artificial neural networks with a correlation coefficient of 0.93 give better results. With the study, different input sets and different algorithms were compared and their effects on output values were examined.