Bitcoin Transaction Fee Estimation Using Mempool State and Linear Perceptron Machine Learning Algorithm

Abstract

Bitcoin, the world’s most valued cryptocurrency, uses a network of computers across the globe to create an immutable transaction record on a public ledger known as the blockchain. The blockchain consists of a series of timestamped blocks, where each block contains a series of transactions selected for inclusion in the block, generally based on how high of a fee the transaction allocates to the party responsible for confirming the transaction. Estimating an appropriate fee for Bitcoin transactions is a challenge for many transacting parties using Bitcoin as a digital currency. This work aims to help Bitcoin users save funds in their transaction fees when building multisig transactions by providing fee estimates that referenced the current state of the unconfirmed transaction pool using the perceptron machine learning classification algorithm. 4