Micro-blog is essentially a kind of web service platform, and it had a wide space with the development of mobile Internet. As an important part of social network, influences among the micro-blog users are becoming a hot spot in the research of the micro-blog. This work has a very important theoretical and practical significance for monitoring public opinion. Through the analysis of user’s behavior patterns using the transfer Entropy theory, this paper set up an improved model for better evaluating micro-blog users’ influences. We processed and analyzed micro-blog users’ behaviors based on time series, and focused on the users’ “tweet”, “retweet”, “comment” and “call”(@) behavior patterns. In turn, this would allow us to gain a better understanding of the characteristics of the new web service platform, and at the same time to find its potential values for researches and applications. To validate our method, we crawled data from the Sina weibo, and these data was processed and analyzed by our method in this paper. The experiment result showed that our method performed well on evaluating users’ influences.