Customer credit evaluation method and device

Abstract

The invention provides a customer credit evaluation method and device. The method comprises the steps of collecting credit evaluation data of a to-be-evaluated customer; performing credit evaluation on the to-be-evaluated customer according to the credit evaluation data and a pre-established credit evaluation model, wherein the credit evaluation model is obtained by performing model training by using input feature data, and the input feature data is determined from preset credit evaluation data for model training, which is subjected to feature selection and discrete processing, according to a preset information threshold value. The invention belongs to the field of Internet of Things, can also be used in the financial field, and provides an effective customer credit evaluation method, aiming at credit business in a bank, the bank risk control capability is effectively improved, the bank auditing cost is reduced, and the risk control efficiency and accuracy of the bank are improved.

Description

technical field [0001] The invention relates to data processing technology, specifically a method and device for customer credit evaluation. Background technique [0002] At present, with the development of economy and society, the Internet financial products of banks are developing rapidly. While providing credit services on the Internet, banks need to quickly and comprehensively review the information of credit customers, so as to reduce the risk of default by customers. In this case, the original manual review method can no longer meet the actual needs. A batch of credit scoring methods trained by machine learning have gradually occupied the market because of their fast, efficient and excellent performance. [0003] In the prior art, most commonly used customer credit scoring methods use a single model algorithm such as SVM, neural network, tree model algorithm, and the like. However, there are some problems in the models trained by these algorithms at this stage. On t...

AI Technical Summary

Problems solved by technology

However, there are some problems in the models trained by these algorithms at this stage. On the one hand, some of the above algorithms generate nonlinear and discontinuous functions to fit user data, which will reduce the interpretability of the model; In the feature engineering stage of the construction process, the useful information in the data is not well explored, resulting in a waste of information, which leads to a decline in the accuracy of model predictions

In addition, the traditional credit scoring method based on a single model cannot fully describe the user's credit picture, and a single model is also lacking in generalization

Images