A comparison between Super Vector Regression, Random Forest Regressor, LSTM, and GRU in Forecasting Bitcoin Price

Rifando Panggabean (Politeknik Caltex Riau)
Yohana Dewi Lulu Widyasari (Politeknik Caltex Riau)

Proceeding International Applied Business and Engineering Conference
Vol. 2 (2022): Proceeding International Applied Business and Engineering Conference 2022

High bitcoin user volume results in high market volatility, and indicators commonly used in stock and forex transactions have low accuracy in handling bitcoin's highly volatile market. The present study aims to find out the most optimal machine learning algorithm for Bitcoin transactions by examining four algorithms: Super vector regression(SVR),Random Forest Regressor(RF),Long short-term memory(LSTM), and Gated Recurrent Unit (GRU), examined using four tests, namely Root Mean Square Error (RMSE), Mean Square Error (MSE) , Mean Absolute Error (MAE) and R-Squared(R2). The test was performed using Bitcoin data between 2014 and 2022. The test result showed that LSTM+GRU algorithm exhibited the highest accuracy, indicated by a R-squared of 94%.