Research On Risk Of Equity-based Crowdfunding Based On Random Forests

As one way of crowdfunding,Equity-based Crowdfunding offers new financing channel for small and medium enterprises,improves cooperative efficiency for financiers and investors,makes financing environment better and provides dynamics for entrepreneurship economy.However,Equity-based Crowdfunding is now at early stage,it still faces multiple risks despite rapid development.Based on what is mentioned above,this paper studies participants of Equitybased Crowdfunding and hopes to offer evidence for risk control of every side by studying risks of pertinent participants.Based on relevant research on Equity-based Crowdfunding,this paper analyzes recent development condition of Equity-based Crowdfunding and finds that participants of every side all have risk problem of different level.First,this paper deeply analyzes four aspects like project creators,project itself,protection measures and subsequent progress by using risk tree searching method,then identifies 30 risk identification indicators of Equity-based Crowdfunding.Second,this paper uses recursive deletion method to do the feature selection for risk identification indicators following the principle of objectivity,scientificity and feasibility.Third,this paper gets nineteen important identification indicators such as Annual Return,Reputation,Local,Engage and Existing Assets and so on,then constructs risk identification system based on feature variables mentioned above.Then this paper constructs Equity-based Crowdfunding classification model by using random forest algorithm and finds random forest model has higher classification accuracy.Fourth,this paper compares random forest model with other classification algorithm model from classification accuracy and error rate's perspective.By comparing classification outcome of random forest with other models,this paper finds that random forest is more accurate in the classification of Equity-based Crowdfunding risk identification,which supports the feasibility of random forest model in the Equity-based Crowdfunding research.At last,this paper offers corresponding risk control measures for project creators,investors and platform based on important indicators such as teammate numbers,annual return,project site,reputation and net margin,offers alternatives for risk control of Equity-based Crowdfunding.