摘要
针对运用信用评分模型提升银行决策能力进行了研究。将支持向量回归模型应用于企业信用评分问题,并提出基于随机子集的支持向量回归集成模型。首先使用随机子集抽样模型获得大量训练数据集,然后使用不同的训练集子集获得差异化支持向量回归模型,最后使用简单平均方法整合不同模型的预测结果。基于企业信用评分数据的实验结果证明了支持向量回归模型的有效性。
This paper researched on using credit scoring models to improve banks＇ decision-making capacity. It applied sup- port vector regression model to the enterprise credit scoring, and then, it put forward a support vector regression integration model which based on random subset. Firstly, it used random subset sampling model to get enough different training data. Secondly, it employed different training subsets to get various support vector regression models. Finally, it integrated the pre- dicted results of different models by using the simple average method. In conclusion, the result of the experiment based on en- terprise credit scoring data proves the effectiveness of the model.
出处
《计算机应用研究》
CSCD
北大核心
2016年第11期3378-3382,共5页
Application Research of Computers
基金
上海市科学技术委员会科研计划资助项目(14511107202
15511107302)
国家自然科学基金资助项目(71101084
71301095)
关键词
信用评分
随机子集
支持向量回归
credit scoring
random subset
support vector regression（SVR）
