Abstract. Mathematical statistics is an important branch of mathematics. It is also an important tool to other sciences, such as social science, management science, etc. With the development of information science and technology, on one hand, the appearance of softwares, such as SAS, SPSS, etc., the tasks of computation which related to statistical analysis can be carried out by softwares, on the other hand, the appearance of big data brings about new challenges for mathematical statistics. What should we learn or teach in the future? In this paper, from the viewpoint of educational innovation, and based on the understanding of big data, we consider the possible reformation on the related methods of mathematical statistics in the future. We intend to provide some suggestions on the methods of statistical analysis which will be furtherly improved or developed, especially the analysis methods for the data with a whole sample, and statistical inferences for figures, sounds and words.