Abstract: Weibo forwarding is an important way to spread information on Weibo. The factors affecting the forwarding of Weibo are mainly user attributes, Weibo content, user social and user interests. The existing forecasting model only considers some factors. In fact, the four factors have an impact on the user's forwarding behavior. In addition, we should also pay attention to the real-time nature of the calculation time of the prediction model. Based on the above analysis, a Weibo forwarding prediction model based on hybrid features and XGBoost algorithm is proposed. Firstly, user features, Weibo features, social features and interest features are extracted according to four factors. Then user influence is calculated based on PageRank algorithm, interest similarity is calculated based on Latent Dirichlet Allocation（LDA） model and KL distance and define the calculation formula of user forwarding activity and interaction intensity between users. Finally, the XGBoost algorithm is used to construct the prediction model and perform forwarding prediction analysis. The experimental results show that the prediction method of this paper has a good performance in the evaluation index of accuracy and time, and also verifies the importance and effectiveness of considering four factors comprehensively.