Abstract:
Considered by many to be the forerunner of smart contract technology, Ethereum has
become a widely used blockchain platform. Beyond being a cryptocurrency, Ethereum's
main purpose is to offer a decentralized platform for smart contract execution. There are
several benefits to using machine learning to predict Ethereum prices in the ever-changing
cryptocurrency markets. Machine learning algorithms are able to identify complex patterns
and trends that may be difficult for humans to analyze by examining large amounts of
historical data. A thorough analysis of Ethereum's price fluctuations is made possible by
this data-driven strategy, which considers market sentiment, technical indications, and
outside events. In this study, I examine historical Ethereum data that I gathered from the
Yahoo Stock Market between 2017 to the present. I used the regression model and the
neural network model as my two sorts of algorithms. I utilize the Huber Regressor, Least
Angle Regression, Linear Regression, Orthogonal Matching Pursuit, and Lasso Least
Angle Regression for my regression models, and I use LSTM for my neural network model.
According to my research, the Huber Regressor performs best when taking R2 score into
account, having the greatest R2 score of 0.9906. LSTM, however, handles errors less
frequently. 35.9866 is the Mean Absolute Error (MAE), which is less than that of any other
model. High accuracy is correlated with low error. The Accuracy of LSTM is 98.3582%
which is higher when taking MAE into account. According to this research, the best model
for high-volume timestamp data is an LSTM.
