A Study and Experimentation on Bitcoin Price Forecasting with Time Series Analysis and Recurrent Neural Network (RNN)

Bitcoin is a peer-to-peer payment system and a digital currency as well as a cryptocurrency. The creation and transfer of money are controlled by cryptography. It is shared through a growing list of blocks linked through a network called the blockchain. The price of bitcoin varies from one day to another, and from the literature, it has been found that there are many fluctuations in its rate. Moreover, many models have been developed to predict the price of bitcoin. This research aims to develop a time series model for solving the problem of bitcoin price prediction. Time series forecasting is a widely applied model, and it learns to predict the future from past performance of series of data. It integrates Auto-Regressive Integrated Moving Average (ARIMA) and recurrent neural network (RNN) to improve the prediction accuracy. In this model, recurrent loops in the activation function of the RNN model are added with ARIMA. The proposed model predicts the bitcoin price with higher accuracy compared with existing models.