Effective GasPrice Prediction for Carrying Out Economical Ethereum Transaction

Abstract

In Ethereum, reaching a transaction consensus costs a certain number of gases, which should be purchased by users in their self-defined gas prices. Generally, the higher the gas price, the shorter the time is spent on reaching consensus. Since the transaction gas prices still vary greatly in a block, generating a reasonable price that can make a trade-off between the consensus time and the gases cost is of great significance. In this paper, we propose a Machine Learning Regression-based gas price predicting approach (MLR), aiming to find the lowest transaction gas price in the next block for carrying out economical Ethereum transaction. Specifically, we identify five influencing factors (i.e., difficulty, block gas limit, transaction gas limit, ether price, and miner reward) from the Ethereum transacting process and resort the classic machine learning regression to build the predicting model. Our empirical study on 194,331 blocks implies that the proposed MLR approach works well and can save $17,552.2 for all transactions in the 74.9% accuracy.