一种新的应用于中国P2P网络借贷行业的q值权重混合机器学习技术

We often encounter imbalanced data in credit risk when there is an unequal representation in the classification categories.In order to provide loan company with a simple and novel approach to get the customer's credit prediction result. An analytic model for machine to learn from data ,then it can be able to do predictive analysis. Here, a machine learning model is needed to build to help the P2P lending sector which sometimes faced with risk challenge when advancing loans to customers. Obviously, the accuracy of the model plays a very important role when the loan companies make decision. The accuracy can be improved by many factors, some of these the use of better machine learning model and balanced data. In this work, we formulate a novel q-weighted hybrid model to gain performance improvement and to solve the problem of imbalanced data and missing value. The machine learning and statistical techniques can be combined in various ways for creating the effective hybrid models. Many Support Vector Machine(SVM) mod- els are combined by a q-weighted hybrid method, and the training sets used to construct the SVM model contained only selected attributes and were composed only of the complete examples. The final classification is made by the test statistic which is sequentially obtained from models. The test statistic are compared with two thresholds to get decision in the majority voting process. The results of the single and hybrid models shows that the proposed hybrid method had the best result.