摘要
提出了一种新的经济效益评价模型:核主成分分析(KPCA).它通过一个非线性变换,将原变量空间映射到高维特征空间,然后在高维特征空间中进行线性主成分分析.通过核技巧,只需在原空间进行点积运算,便可使第一主成分的贡献率达到90%以上,能有效避免PCA中因各指标贡献率过于分散而影响评价效果.将该模型应用到广东8个卷烟企业进行评价,得到了较理想的评价效果.
A new kind of comprehensive evaluation model (kernel principal component analysis) is proposed in this paper, it can effectively compute principal component in high dimensional feature spaces by using kernel function, but we need not know the form of nonlinear transforming. We can make the first principal component contribute to 90% by choosing the suitable kernel function and point product computation only. Finally we can achieve more objective and reasonable evaluation results while it was applied to evaluation about eight enterprises in GuangDong.
出处
《数学的实践与认识》
CSCD
北大核心
2006年第1期35-38,共4页
Mathematics in Practice and Theory
基金
国家自然科学基金项目(60573158)