A Comparative Study to Forecast Bitcoin Price Using Machine Learning & Time Series Models
Authors:-Pasumarthy Manasa Komali, Ruma Dutta
Abstract-Block chain is an emerging technology and cryptocurrency is one of the backbone of blockchain technology in the financial sector. Even though other cryptocurrencies have been around for a few years, Bitcoin has kept its position as the most valuable cryptocurrency. On the other hand, Bitcoin’s price has been very unpredictable, making it almost impossible to guess where it might go in the future. The goal of this research is to find out which model is the most accurate and best at making predictions. Several different machine learning and time series approaches are used to determine Bitcoin prices. The goal of this article is to see how well it is possible to predict where the price of Bitcoin in US Dollars will go in the future. US Dollars are used to figure out how much Bitcoin costs. The price information came from a website called Coin market cap. We used machine learning techniques like Random Forest, XG Boost, kNN and time series models like AR, MA, ARIMA, Prophet. It has been found in our experiments that time series models give better results than machine learning approaches.
