Aprendizado de Máquina para a Previsão do Comportamento de Preços da Criptomoeda Ethereum

Abstract

Development of machine learning algorithms for the prediction of the behavior of prices of the cryptocurrency Ethereum using a public and widely available database (Kaggle). One linear model was developed (ARIMA, time series analysis), while three non-linear models were developed (deep learning models, using long term short memory framework). As a result, it was noticed that one of the deep learning models was able to predict prices that were, in average, only 4,32% far apart from real prices.