首页> 外文期刊>Statistics and computing >Robust nonlinear principal components
【24h】

Robust nonlinear principal components

机译:鲁棒的非线性主成分

获取原文
获取原文并翻译 | 示例
       

摘要

All known approaches to nonlinear principal components are based on minimizing a quadratic loss, which makes them sensitive to data contamination. A predictive approach in which a spline curve is fit minimizing a residual M-scale is proposed for this problem. For a p-dimensional random sample x_i (i = 1,..., n) the method finds a function h: R → R~p and a set {t_1,..., t_n) is contained in R that minimize a joint M-scale of the residuals x_i - h(t_i), where h ranges on the family of splines with a given number of knots. The computation of the curve then becomes the iterative computing of regression S-estimators. The starting values are obtained from a robust linear principal components estimator. A simulation study and the analysis of a real data set indicate that the proposed approach is almost as good as other proposals for row-wise contamination, and is better for element-wise contamination.
机译:非线性主成分的所有已知方法都基于最小化二次损失,这使其对数据污染敏感。针对此问题,提出了一种预测方法,其中拟合样条曲线以最小化残留M比例。对于一个p维随机样本x_i(i = 1,...,n),该方法找到一个函数h:R→R〜p,并且集合{t_1,...,t_n)包含在R中,该函数使a最小化。残差x_i-h(t_i)的联合M尺度,其中h处于具有给定结数的样条族上。然后,曲线的计算成为回归S估计量的迭代计算。起始值从稳健的线性主成分估计器获得。仿真研究和对真实数据集的分析表明,对于行污染,该方法几乎与其他方法一样好,对于元素污染,该方法更好。

著录项

  • 来源
    《Statistics and computing》 |2015年第2期|439-448|共10页
  • 作者单位

    Faculty of Exact Sciences, University of La Plata, C.C. 172, 1900 La Plata, Argentina;

    Faculty of Economic Science and Statistics, University of Rosario, Bv. Orono 1261, 2000 Rosario, Argentina;

    Departamento de Matematica, Faculty of Natural and Exact Sciences, Universidad de Buenos Aires, Ciudad Universitaria, Pabellon 1, 1428 Buenos Aires, Argentina;

  • 收录信息
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类
  • 关键词

    S-estimators; Splines; Principal curves;

    机译:S估计量;花键;主曲线;

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号