基于灰狼优化算法和最小二乘支持向量机的信用评估  

Credit evaluation for hybrid grey wolf optimization and least squares support vector machine approach

作　　者：周敏[1,2] ZHOU Min(School of Computer Science,Civil Aviation Flight University of China,Guanghan 618307,China;School of Economic Information Engineering,Southwestern University of Finance and Economics,Chengdu 610074,China)

机构地区：[1]中国民用航空飞行学院计算机学院 [2]西南财经大学经济信息工程学院

出　　处：《成都理工大学学报（自然科学版）》2019年第4期507-512,共6页Journal of Chengdu University of Technology: Science & Technology Edition

摘　　要：最小二乘支持向量机(LSSVM)被证实是一种有效的信用评估方法,然而,传统的交叉验证和网格方法通常得不到最优参数。为了解决这个问题,作者提出了一种改进的灰狼优化算法(IGWO),该算法能非线性地调整收敛因子,并能自适应调整α狼、β狼和δ狼对ω狼的影响。然后,提出了一种用IGWO来优化LSSVM参数的方法IGWO-LSSVM,并将其应用于信用评估中。在公开的德国和澳大利亚真实信用数据集上,IGWO-LSSVM较传统的K近邻、朴素贝叶斯、决策树、支持向量机和LSSVM等信用评估方法均有明显的提升,表明IGWO- LSSVM 是一种有效的信用评估方法。The least squares support vector machine (LSSVM) has been proved to be a powerful tool for credit evaluation. However,traditional parameter tuning methods such as cross validation and grid search can not obtain global optimal parameters for LSSVM. In this paper,an improved grey wolf optimization (IGWO) and LSSVM approach (IGWO-LSSVM) for credit evaluation is proposed so as to solve the problem. It shows that the IGWO nonlinearly updates the convergence factor and adaptively adjust the impacts of α,β,and δ wolves on ω wolf. The IGWO is firstly used to search global optimal parameters for LSSVM and then LSSVM was applied to the credit evaluation. Experimental results on two real-word credit datasets (German and Australian) show that the proposed IGWO-LSSVM outperforms the competing state-of-the-art methods on credit evaluation. Therefore,the proposed IGWO-LSSVM is an effective method for credit evaluation.

关 键 词：灰狼优化算法 最小二乘支持向量机 信用评估 金融风险 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] F830.5[自动化与计算机技术—控制科学与工程]