Bitcoin is a decentralized electronic cryptocurrency, and the anonymity of the transaction address hides the real identity of the transaction user, leading to Bitcoin being used by some unscrupulous elements in various illegal activities. By analyzing the transaction attributes and behavioral characteristics of known entities, the category prediction of transac⁃ tions of unknown entities can be performed by using machine learning methods. First, this paper outlines the sources of bit⁃ coin entity categories and classification labels; second, it analyzes and summarizes the machine learning-based bitcoin enti⁃ ty classification methods; finally, it analyzes the problems faced at this stage and predicts the future development trends.