Analysis of Bitcoin Price Prediction Using Machine Learning

The research purpose of this paper is to obtain an algorithm model with high prediction accuracy for the price Bitcoin on next day through random forest regression and LSTM, explain which variables have influence Bitcoin. There much prior literature research, methods mainly revolve around ARMA time series LSTM deep learning. Although it cannot be proved by Diebold–Mariano test that significantly better than errors RMSE MAPE are those LSTM. changes in determine each period also obtained regression. From 2015 2018, three US stock market indexes, NASDAQ, DJI, S&P500 oil price, ETH impact prices. Since important become Japanese index JP225. relationship between number periods explanatory brought into shows predicting day, only one lag has best accuracy.