Bitcoin Price Prediction Combining Data and Text Mining

Abstract

The idea implemented in this paper is the creation of a machine learning model, which utilizes past prices of Bitcoin, Google trends data and some custom features, which were created by text mining on tweets about Bitcoin. The aim of this study was to predict future Bitcoin prices. For this purpose, we compared a Deep Neural Network, and specifically, a network with LSTM layers, a Gradient Boosting Regression model and an XGBoost model. According to our results, the Deep Neural Network performed better, with a significant root mean square error of 0.999% on test data.