Prediction of Ethereum Prices Using Linear Regression and Long Short-term Memory

Abstract

Contemporarily, the popularity and use of cryptocurrencies has risen along with their prices and Ethereum is the second most popular and largest cryptocurrency after Bitcoin. Cryptocurrencies are based on the blockchain, which is a decentralized technology that has the power to change any banking system. They have become an attractive investment for both traders and individuals looking to invest. The price of Ethereum fluctuates and is affected by various factors, e.g., the crypto trading exchange as well as supply and demand. Ethereum is so valuable because it can be used as cash and one also pay Ethereum in full or in part to someone in exchange. Besides, it is easily guaranteed by the blockchain. Unlike stocks, the price of Ethereum is much more variable because it is traded 24 hours a day and there are no closing times. On this basis, this paper compares the results of two different models, namely linear regression and Long Short-Term Memory networks (LSTM). The dataset comprised in the closing prices of the last 372 days for Ethereum. The performance of the obtained models is critically evaluated using statistical indicators Root Mean Squared Error (RMSE) and the study have drawn our conclusions based on the RMSE result. The paper demonstrates a technique for using time series data in both models and determining each model's RMSE. These results shed light on guiding further exploration of prediction Ethereum prices and trends.