Crypto- Currency Price Prediction with Decision Tree Based Regressions Approach

Document Type : Research Paper


1 Department of Statistics, Allameh Tabatabyee University

2 Iran Banking Institute, Central Bank of Iran


Generally, no one can reject the fact that crypto currency market is expanded rapidly during last few years as, nowadays, crypto currency market is attractive for both traders and business who are not willing to pay for FATF services for transferring money. With this in mind, crypto currency price prediction is crucial for many people and business entities. While there have been quite a few conventional statistical models to forecast crypto currency prices, we decided to make price prediction using decision Tree Based Regression. In this research we devised a decision tree models to predict Bitcoin which is the most renowned and frequently used crypto currency. we used Volume from, Volume to, New addresses, Active addresses, large transaction count, Block height, Hash rate, Difficulty, Current supply as predictor variables in addition to historical crypto currency price data during the with a total of 1000 Observations. We find that forecasting accuracy of decision tree models are higher than benchmark models such as linear regression and autoregressive integrated moving average(ARIMA).