Risk assessment method and device and storage medium

Abstract

The embodiment of the invention relates to the field of risk control. Some embodiments provide a risk assessment method and device and a storage medium. The method comprises the steps of obtaining user data of a to-be-evaluated new application user; extracting a first backtracking feature of the user data; comparing the first backtracking feature with each second backtracking feature and each third backtracking feature of a preset sample group to obtain similarity between the first backtracking feature and each second backtracking feature and similarity between the first backtracking feature and each third backtracking feature; and determining a risk assessment result of the new application user according to the similarity. According to the embodiment of the invention, the screened target rejection sample and the unbiased model obtained through sample fusion training are used as the risk assessment model, so that the default probability of the new application user is calculated through the risk assessment model, and the assessment accuracy of new user default risk assessment is effectively improved.

Description

Technical field [0001] The present application embodiments relate to the field of wind control, and more particularly to a risk assessment method, device, and storage medium. Background technique [0002] In the wind control modeling scene, it is mainly to predict the applicant's default risk by building a model, thereby giving the applicant's default probability to assist judgments whether loan is issued. [0003] In the relevant technique, the evaluation model performance in the wind control construction field scene needs to consider the actual performance effect after the model is online. However, it is often a case where the sample deviation has an overestimation model, resulting in a superior performance of the model in the test data set, but the effect is not expected after the actual online. The reason for forming a sample deviation is that at the statistical analysis, the overall sample is replaced with a partial sample, and whether the partial sample is sufficiently repr...

AI Technical Summary

Problems solved by technology

Although the method of rejecting inference can bring some benefits to the problem of sample deviation, the rejecting inference relies on the assumption that the model passing through the customer group is effective in rejecting samples, so the model is often not effective in rejecting samples, resulting in the final The effect of rejecting the inference is not as expected

Images