BITCOIN PRICE PREDICTION WITH RANDOM FOREST REGRESSION ALGORITHM

Abstract

With the rapid developments in today's technologies, people can now perform their payment and shopping transactions through digital platforms. However, payment security problems in e-services have led people to seek alternative payment methods. Thanks to blockchain technology, cryptocurrencies that are not dependent on the central authority and can be paid in a completely secure way have been developed. Bitcoin is a digital currency that is not tied to a central authority or bank, introduced in Satoshi Nakamoto's 2008 article entitled "Bitcoin: The Peer-to-Peer Electronic Money System". Bitcoin, which attracts the attention of investors in the financial world, especially during the pandemic process, is traded in a market with high volatility. For this reason, it is of great importance for those who want to make forward price predictions. In this study, it is aimed to develop a price prediction method that will contribute positively to the profit share of Bitcoin investors. With Bitcoin, the data belongs to a time series, and Random Forest Regression, a model used to predict time series, was used. The model is trained on two years of Bitcoin data for the years 2020-2022. The statistical error measures of the model were calculated as MSE, R2, MAE and RMSE as 0.031%, 99.39%, 31.16% and 55.33%, respectively.