摘要
针对数据预处理的方法进行了研究,提出了基于非线性相关性分析与量化(Non-LinearCorrelation Analysis,NLCA)算法。NLCA算法是一种基于在多重图中通过对多重边聚合从而达到约简的工具,它包括边聚合与点聚合。这种算法能够很好地表示实时数据全局的相关性,改进了现有使用联合概率的单一计算方法。对该算法进行了大量实际数据的验证,显示出它是一种优于现有的数据预处理方法。
An estimation model for credit risk and the optimal portfolio in the commercial bank has become an extremely important subject to business dealing. A novel pretreatment method of Non-linear Correlation Analysis(NLCA) was presented in the data mining of credit risk evaluation and the optimal portfolio, and the aggregation character of the conjunction graph was studied , then the convergence and validity of the arithmetic of NLCA was put out , at last applying it to the real data.
出处
《计算机应用》
CSCD
北大核心
2006年第6期1406-1408,共3页
journal of Computer Applications
基金
广东省科技基金项目(2005B101010332004A10202001)
关键词
非线性相关性分析
数据挖掘
信用风险评估
聚合
关联图
Non-Linear Correlation Analysis(NLCA)
data mining
credit risk ewfluation
aggregation
conjunction graph