摘要
针对一类高维小样本数据,利用统计方法的非参数检验与偏最小二乘回归(PLS)构造小样本预测模型,实现基于Wilcoxon秩和检验的变量选择与基于PLS的变量压缩降维.并通过DNA序列分类问题实现基于统计方法的小样本数据建模与可视化,计算结果表明方法对小样本具有可行性、有效性.
With a class of high-dimensional ＆ sample are constructed using Hypothesis Test statistical methods,which carried out variables small-sample data, Prediction models of small- and Partial Least-Squares Regression（PLS） of selection on wilcoxon rank-sum test and variables compression ＆ dimension reduction on PLS. And small-sample modeling and visualization on statistical methods are achieved by the instance of DNA sequence classification, the results show that the modeling methods of small-samples have feasibility and stability.
出处
《数学的实践与认识》
CSCD
北大核心
2013年第7期68-75,共8页
Mathematics in Practice and Theory
基金
高校博士点专项科研基金(20070384003)
福建省教育厅科技项目(JB08244)
