摘要
提出了一种改进的支持向量分类方法——NN-SVM-KNN。用最近邻((NN)对训练集进行修剪,用支持向量机(SVM)进行分类,同时在分类过程中,判断待识样本是否落在间隔之外,若是直接用SVM分类,否则,用所有的支持向量作为训练集,进行K近邻分类。并把这种方法应用到个人信用评估中,和其它信用评估方法比较,得到了比较好的结果。
A new method of support vector machine(SVM) ──NN-SVM-KNN is presented. It first prunes the training set using NN, then classifies the new training set with SVM, in the classifying process, tests whether the test sample is in the margin, the tested sample would be classified with SVM. Otherwise, the KNN method will be used. Furthermore, this new method is applied to personal credit scoring, and it reports better results when contrasted with the other methods of credit scoring.
出处
《计算机工程》
CAS
CSCD
北大核心
2005年第8期153-154,共2页
Computer Engineering
基金
国家自然科学基金资助项目(10371131)
关键词
支持向量机
信用评估
最近邻
Support vector machine
Credit scoring
Nearest neighbor(NN)