IEEE International Conference on Electrical, Automation and Computer Engineering (ICEACE)
Optimization of Credit Score Card Portfolio Based on QUBO Model with Quantum Annealing Algorithm
Siqi Feng, Jinfeng Shi, Rui Zhang, Yuyang Wang, Fan Chen, Xu Wang
Abstract
With the progress of science and technology, quantum computing plays an increasingly important role in various fields, solving the traditional arithmetic bottleneck. In the financial field, credit scoring is the core of the lending industry and one of the most common risk management tools. Researchers are trying to find the best threshold-setting scheme to maximize revenue through quantum computing. This approach not only solves the optimization problem in the field of credit score cards, but also provides a new way of thinking about optimization problems in other fields. In this paper, three traditional computational models, one-fold credit card strategy, two-fold credit card combination strategy, and three-fold credit card combination strategy, are proposed, and the credit score card scheme that maximizes the final revenue is solved for each of the three models. The results show that under the one-fold credit card strategy, the final income is maximized at $54,087.2; under the two-fold credit card combination strategy, the final income is maximized at $41,106.3; and under the three-fold credit card combination strategy, the final income is maximized at $27,914.8. Secondly, this paper converts these three strategies to the QUBO model and obtains the maximum final income that is consistent with the maximum under the traditional model, with an accuracy of 96.3%, 92.7%, and 89.9% for the three combination strategies, respectively.