In the era of big data, more and more elements of data visualization have emerged in the rapidly changing information journalism, and visualization technology has taken on an increasingly important role in journalism. In this paper, 50 representative data journalism projects using visualization are compiled from actual cases of data journalism, and a classification framework is proposed based on both theory and practice of data journalism. It helps to analyze the role of visualization at 2 levels: production and understanding of data news. In the process of data news production, the division of roles, types of visualizations, and production techniques in different news processes are discussed; in the process of user understanding of data news, the characteristics of data news that facilitate understanding, user interaction patterns, and user perceptions are discussed. The proposed framework helps to better understand the process of combining data journalism and visualization, and summarizes the issues and working patterns within the corresponding framework. The paper concludes with a discussion of the direction of visualization in data journalism, where people can interact with visualization through virtual reality and other technological means in new communication methods, and take advantage of the opportunities of automatic generation technology; at the same time, they face some problems and challenges, such as weighing the complexity of interaction; the paper also suggests some suggestions for data journalists to integrate data journalism with visualization.