The characteristics of the era of big data have brought about changes in various aspects of college English writing teaching, from concept to behavior, and have also brought new opportunities and challenges to the reform of college English writing teaching. In this paper, we start by reforming college English writing teaching, imposing constraints on simplified information forms, and constructing the SVAR model to explore the intrinsic connections between English writing teaching data. Finally, it examines the causes of the errors that impact college students’ English writing process and proposes measures to decrease the number of writing errors. The experimental results show that the sample test t-value of English writing pre and post-test scores of the experimental class and the control class are 4,978 and 4.905, respectively, and the Sig. The values are 0.000 and 0.245, respectively. Only the experimental class shows a significant difference, and the reform of college English writing teaching has been effective.