信用评估中的鲁棒赋权自适应L_p最小二乘支持向量机方法

A Robust Weighted Adaptive LpLS-SVM Method for Credit Risk Assessment

消费者信用评估是金融风险管理和信用产业竞争的一个重要方面。信用评估数据中常带有噪声点,并且其类别是不均衡的。最小二乘支持向量机是一个被广泛应用的分类模型,其模型简单,求解速度快,但鲁棒性差。本文提出了一个鲁棒赋权自适应Lp最小二乘支持向量机模型,能够适应信用评估样本数据库类别不均衡的特点,可以有效处理信用评估数据中带有噪声点的问题。在仿真数据和三个信用数据库上的实证分析表明,本文所提出的模型具有较好的鲁棒性和分类能力。

Consumer credit risk assessment is an important aspect of financial risk management and credit industry competition.Credit database often contains noisy data,which makes the data uncertain.Least squares support vector machines,a widely used binary classification model,is simple and easy to be applied.In this paper,we propose a robust weighted adaptive Lp least squares support vector machines,which can deal with unbalanced data sets and noisy data.The empirical test on simulation and three credit data sets have shown the model has outstanding robustness and generalization ability.