Bitcoin Price Prediction- an Analysis of Various Regression Methods

Abstract

Cryptocurrency is a fascinating area of research developed due to the rapid development of financial technologies. One of the well-received cryptocurrencies is Bitcoin. The motivation behind this paper is to predict bitcoin prices with high accuracy using various regression-based models. Bitcoin is the fastest growing cryptocurrency. We have seen drastic changes in bitcoin prices over time. To make predictions for such changes, we use machine learning techniques over real-time data recorded for every 24-hour time interval since the presence of bitcoin beginning in the year 2009. We select eleven different regression models, analyze these models and obtain the best regression-based model for bitcoin price prediction. The results obtained from the study depict that the Bayesian ridge regressor outperforms all the other regression-based models followed by the Linear regressor.