Credit risk control model generation method, credit evaluation method and system, machine readable medium and equipment

Abstract

The invention provides a credit risk control model generation method. The method comprises the steps of obtaining one or more combined features corresponding to original attribute data of a credit business object; determining segmentation points and segmentation intervals according to the feature conditions of the combined features; determining a new feature condition according to the segmentationpoint and the segmentation interval; and performing training according to the combined features and new feature conditions corresponding to the combined features to obtain a credit risk control model. According to the credit risk control model for generating the second-order feature combination score card based on the gradient boosting tree, business personnel can clearly master details in the decision process of the model and can introduce business knowledge to check and finely adjust each detail, so that the requirement of credit risk control business for model interpretability is met.

Description

technical field [0001] The present invention relates to the field of credit risk control, in particular to a method for generating a credit risk control model, a credit evaluation method, a system, a machine-readable medium and equipment. Background technique [0002] In the field of credit risk control, the traditional scorecard model based on logistic regression is traditionally adopted due to business requirements for model interpretation. In the training phase, each feature column of the input data is binned, and the score of each binning interval is calculated through the training data to obtain a score card. In the prediction stage, the scores will be calculated by comparing each column of the input data against the scorecard, and finally summed to obtain the predicted credit score. [0003] However, since logistic regression is a linear model, the traditional scorecard model suffers from two major flaws. [0004] First of all, the impact of certain variables on cred...

AI Technical Summary

Problems solved by technology

Therefore, the scorecard model has strong requirements for the individual predictive ability of each input feature, and cannot make full use of the combination of some weak features to improve the predictive ability

[0006] These two defects have led to the feature selection, feature engineering and binning steps of the scorecard model relying on modeling experience, expert knowledge and cumbersome manual parameter adjustment. Failure to do any of the steps will lead to the model not performing as expected.

Images