Research and Empirical Analysis on the Optimal Combination Mechanism of Credit Scorecard Based on QUBO Model

Abstract

This paper selects relevant specific data for empirical analysis. By randomly crawling the background network data of a bank and according to the current relevant national financial policies and plans, 100 reasonable scorecards are formulated. Then, the QUBO model is innovatively introduced to optimize the combination problem of bank credit scorecard: The integer programming operational problems are constructed, and the variables in them are transformed into 0-1 binary decision variables. The data information is represented by the mathematical expression of Hamiltonian energy, and the QUBO matrix in the standard form is obtained after the transformation of constraints. Finally, the solver in the OPTI Toolbox of matlab is used to solve the problems. Find the scorecard that maximizes the final income and its corresponding unique threshold, and finally calculate the maximum benefit value. The QUBO credit card optimization model established in this paper is highly applicable, covers a wide range of industries, and has accurate prediction value, which maximizes the risk of bank financial collection loss, and provides a feasible and effective scheme for the financial industry.