The traditional database technology can hardly complete full data analysis on mass accumulated attendance data for the enterprise over the years, while big data technology is regarded as a better way to achieve it. After the study on a data-driven attendance management system, this paper proposed a decision tree regression analysis method based on the Spark platform, analyzing mass attendance data in a short time and generating a decision tree to show what affects employee attendance. The decision tree has certain reference significance for the monitoring and warning of employee attendance and employee management decision. This paper detailed the process of dataparallel processing and the regression analysis method of employee attendance by the Classification and Regression Tree (CART) algorithm and compared the efficiency of traditional database technology and parallel data processing for employee attendance data analysis, which verified the effectiveness of the method.