The Application Of The Integrated Model Based On The Voting Classifier In The Identification Of Enterprise Dishonor Problem

It is an important guarantee to standardize the market economy and the premise to construct a good economic and trade environment to identify the problem of enterprise dishonesty.The classification problem of general research is some data sets with balanced distribution of categories,but the problem of untrustworthy enterprise recognition is two data sets with unbalanced distribution of data,so the result of the model prediction of training will be more inclined to the one with more category data.Dishonest enterprise data in order to solve the centralized type distribution imbalance,improve the model of enterprise is faithless recognition correct rate,this paper introduced a sample to increase the small categories the characteristic information of the data contained in the training model of data need to the two kinds of a balance,again on the basis of introducing the voting mechanism of learning classes are combined,make the child learning formed complementary advantages and disadvantages between the new model.In addition,a variety of other models are cited to study the data set,and then all models are evaluated according to the unified standard.The accuracy of the model and the type 1 / type 2 error rate are compared as evaluation criteria,indicating that our new model based on the voting mechanism has better classification robustness.