Since the Chinese government continuously support the development of new energy vehicles (EVs), the charging process of EVs will generate big data regarding the EVs charging behavior.This paper proposes a big data mining technique based on Random Forest (RF)and Principle Component Analysis (PCA)for EV charging behavior to identify and analyze clusters with different charging characteristics.Then, Dundee′s EV charging data in the January of 2018 is applied to conduct experiments, and respectively obtains the charging behavior clusters of the workdays, weekends and holidays.Finally, the RF algorithm in the EV clustering problem is compared to the Euclidean distance method and the clusters obtained by RF get more convinced characteristics.