Characteristic factor determination method and equipment

Abstract

This embodiment provides a method and device for determining characteristic factors. The method includes: first obtaining a reference AUC value based on N candidate characteristic factors, each of which is used to describe a type of risk control characteristic, and the The type of risk control features includes at least one of the following: insurance features, underwriting features or claims features; then according to the importance of each of the candidate feature factors, obtain the AUC critical value and AUC fitting curve; then according to the benchmark AUC Value, the AUC critical value and the AUC fitting curve, determine the target eigenfactor in the N candidate eigenfactors, overcome the use of single-round knife-cut method to screen eigenfactors, and cannot evaluate the number of eigenfactors entering the model , the number of eigenfactors entered into the model can only be estimated by the staff themselves, there are deviations, and the best eigenfactors cannot be reasonably selected.

Description

technical field [0001] The invention relates to the technical field of insurance risk control, in particular to a method and equipment for determining characteristic factors. Background technique [0002] At present, there are some fraudulent behaviors in accident insurance and health insurance. For example, some policyholders fill in false high-income information on the application form, and use this information to obtain high insurance coverage, causing great suspicion of fraud. For such behaviors, insurance business scenarios can be used to build a comprehensive, systematic, and business-suitable mathematical model, and then combined with the business scenario rule engine to screen false information in multiple dimensions and apply it to underwriting rules. to avoid fraud. [0003] At present, there are generally three types of optimization of mathematical models: algorithm optimization and sample optimization, that is, to select a part of high-quality samples and optimi...

AI Technical Summary

Problems solved by technology

[0005] The embodiment of the present invention provides a method and equipment for determining characteristic factors, which overcomes the need to use the single-round knife-cutting method to screen characteristic factors, and cannot evaluate the number of characteristic factors entered into the mold. The number of characteristic factors entered into the mold can only be determined by the staff themselves Estimation, there is a deviation, and it is impossible to reasonably select the best eigenfactor

Images