A Review on Machine-Learning Techniques for Estimation of Bitcoin Pricing

After the incorporation of machine learning algorithms and regression techniques, this research exhibits the comparison between their individual accuracies. It uses regression models with 80 percent training and 20 percent testing allocation correspondingly with a random state of 10. Various models such as linear regression, lasso regression, ridge regression, polynomial regression, k-nearest neighbour and support vector machine (SVM-Linear) are used for comparison of their accuracies. The obtained dataset contains a variety of parameters which are considered for the comparison and thereby reviewed.

I. Introduction

Bitcoin is a type of crypto currency that originated in January 2009. It was based on the research and idea of a recluse Japanese individual named Satoshi Nakamoto [1]. The details of the individual or the organization that developed this cutting-edge technology is still unknown. Bitcoin promises lesser transaction fees than conventional online payment methods and, differs from government-regulated currencies, as it is employed by a devolved authority consisting of the entire blockchain [2].