Internet Financial Personal Credit Assessment Based On Support Vector Machine (SVM) Method

In recent years,with the fast development of China's economy,the credit consumption is increasing rapidly,and the size of a variety of individual consumption credit such as housing mortgage,credit card consumption,car loans,and so on are appearing the trends of rapid expansion.All commercial banks begin to take developing individual consumption credit as an important part of their future development strategies.But for the present,the risk management level for individual consumption credit of domestic commercial banks is still relatively low,the management method is relatively backward,and there still lacks a set of scientific personal credit evaluation system.The general ability of model will directly affect profit of credit institution,which demands credit institution possess a scientific and perfect credit management system,thus the study on personal credit evaluation is of high significance.Besides,considering that under big data environment,the size of data is increasing rapidly.In the period with massive data resources,the credit rating industry are faced with both new opportunities and new challenges.In response to the problem of personal credit evaluation under the era of big data,this paper undertake exploration on data analysis and integration.Credit model is the main tool for credit risk assessment.Most quantitative methods have been widely used for credit scoring in finance and banking.Support vector machines are a set of data-driven,supervised learning methods that do not require specific assumptions on the underlying data generating process.This feature is particularly appealing for practical business situations where data are abundant or easily available,even though the theoretical model or the underlying relationship is unknown.In most practical applications,SVM generalization performs significantly better than competing methods.This paper proposes an ensemble model,called RSBC-SVM,which is based on three popular ensemble strategies,i.e.,bagging,random subspace and uses SVM as base learner.Both theoretical and experimental researches show that combining a set of generalization ability and diverse classifiers will lead to a powerful classification system.For the first condition,generalization ability,we choose SVM as the base learner and apply pattern search to find optimal parameters.And for the diversity,among the diverse ensemble methods that are available,bagging and random subspace are two more often used methods and have been found to be accurate,computationally feasible across various data domain.Diversity in bagging is obtained by using bootstrapped replicas of the training dataset: different training data subsets are randomly drawn-with replacement-from the entire training dataset.Each training data subset is used to train a different base learner of the same type.In the random subspace,the training dataset is also modified as in bagging.However,this modification is performed in the feature space(rather than instance space).The random subspace may benefit from using both random subspaces for constructing the base learners and aggregating the base learners.In the model selection,a new reliability-based ensemble strategy with mining correlations among classifiers is proposed for SVM ensemble models.The models were compared with 5 single methods on the basis of tests on real-world credit datasets.The results show that ensemble models have good generalization ability.