Prediction of Bitcoin Price using Data Mining
Dharminder Singh Virk · TRAP@NCI (National College of Ireland) · 2017
Bitcoin is a computerized digital money and exchange network, represents an essential change in financial sectors, an interesting number of customers and excellent evaluation of channel inspection. In this research, dataset related to ten cryptocurrencies are used and created a new dataset by taking the closing price of each cryptocurrency for the research goal to ascertain how the direction and accuracy of price of the Bitcoin can be predicted by using data mining methods. Features engineering evaluated that all the ten cryptocurrencies are strongly correlated with each other. The task is achieved by implementation of supervised learning method in which random forest, support vector classifier, gradient boosting classifier, and neural network classifier are used under classification category and linear regression, recurrent neural network, gradient boosting regressor are used under regression category. In the classification category, support vector classifier achieved the highest accuracy of 62.31% and precision value 0.77. In regression category, gradient boosting regressor got the highest R-squared value 0.99.