Stable variable confirming method and device, server and storage medium

Abstract

The embodiment of the invention discloses a stable variable confirming method and device, a server and a storage medium. The stable variable confirming method comprises the following steps: selectinga first dependent variable and a second dependent variable for a scorecard model by a server; performing independent variable screening on a plurality of candidate independent variables correspondingto the first dependent variable, so as to confirm a target independent variable from the plurality of candidate independent variables; confirming each first weight of evidence WOE value correspondingto each group respectively on the basis of the first dependent variable, and confirming each second weight of evidence WOE value corresponding to each group respectively on the basis of the second dependent variable; and confirming the target independent variable as the stable variable used to establish the scorecard model if each first WOE value and each second WOE value satisfy the stable variable selection condition. By means of the invention, not only the over-fitting of the scorecard model can be avoided, but also the robustness of the scorecard model can be increased, thereby improving the accuracy of the prediction result of the scorecard model.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a method, device, server and storage medium for determining stable variables. Background technique [0002] The scorecard model is a prediction method, especially in the fields of credit risk assessment and financial risk control, which can adjust various parameters of the rating quantification system at any time according to changes in the financial environment and customer credit, establish a reasonable rating system, and guide financial institutions in In the credit business, reasonable ratings are given to relevant customers to avoid credit risks. [0003] The traditional scorecard model will define the dependent variable at the beginning of the modeling, and calculate the weight of evidence (Weight of Evidence, WOE) of the respective variables corresponding to the dependent variable on the basis of the dependent variable, and then determine the stability from the r...

AI Technical Summary

Problems solved by technology

However, the selection method of this stable variable is too single, which makes the established scorecard model prone to defects such as overfitting and poor robustness.

Images