Credit rating optimal index combination selection method based on Informedness coefficient

Abstract

The invention provides a credit rating optimal index combination selection method based on an Informedness coefficient, belongs to the technical field of credit services, and aims to solve the problems that an existing credit evaluation system cannot ensure that the overall default identification capability is strongest, and meanwhile, the correlation between indexes is not considered when a groupof the indexes is selected. The Informedness coefficient-based maximum default identification capability of credit scores is taken as an optimal index combination standard; whether the indexes are selected into an index combination or not is taken as a decision variable; the Informedness coefficient-based maximum default identification capability is taken as an objective function, thereby reflecting that repeated indexes of information cannot be simultaneously selected as constraint conditions to build a 0-1 planning model; and an optimal index combination of credit rating is selected. According to the method, the Informedness coefficient-based maximum default identification capability of the whole credit evaluation system is ensured; and a decision basis of effectively identifying the credit risk is provided for a bank, a credit rating mechanism and the like.

Description

technical field [0001] The present invention provides a selection method for an optimal index combination of a credit rating system, which specifically involves taking the Informedness coefficient of the credit score as the maximum default identification ability as the optimal index combination standard, taking whether the index is selected into the index combination as a decision variable, and using Informedness The method of establishing a 0-1 programming model and selecting the optimal combination of indicators for credit rating belongs to the field of credit service technology, where the coefficient default identification ability is the largest as the objective function, and the information duplication indicators cannot be selected at the same time as the constraint condition to establish a 0-1 programming model. Background technique [0002] Credit is a loan activity conditional on the repayment of principal and interest. The purpose of credit rating is to evaluate the ...

AI Technical Summary

Problems solved by technology

[0007] Existing research has the following problems in constructing the indicator system: On the one hand, the existing research only builds the indicator system from the perspective of whether a single indicator has the ability to identify defaults, and does not take into account the strong identification of individual indicators and the overall default identification of the indicator system. Phenomenon that is not necessarily strong

On the other hand, even if a group of credit rating indicators is selected, the sequence selection algorithm, Lasso algorithm, and stepwise regression method do not consider the correlation between indicators, and it is very likely that indicators that reflect the same information as each other will be selected into the indicator system. Causes redundant information reflected in the index system

Images