In this paper, we employ three classification methods, that is, the Logistic Regression(LR), the back propagation(BP) neural network and the support vector machine(SVM) to predict the result of reward crowdfunding projects. The results show that BP neural network has the best performance. In addition, we use the BP neural network to achieve a reasonable range of financing goal of crowdfunding projects. The experiment shows that the current maximum financing goal for reward crowdfunding is about 1 million RMB, which helps the sponsors to set a reasonable amount of financing target under the current reward crowdfunding market.

In this paper, we introduce a new variable in the model of finding the effect of successful crowdfund the projects. This factor can significantly improve the Pseudo R2 of the evaluation model, which, can be used as the one in predict the result of the future crowdfunding projects.

In this paper, we build the crowdfunding process based on the quality signal and trust theory, and then summarize and select the useful factors by collecting data from famous crowdfunding platforms; then, we empirically analyse the factors that may impact the successful financing of the project. Furtherly, we introduce a new variable to improve the model. The empirical results show that these factors can significantly affect the project’s successful financing.