Explaining a presence of groups in analytical data in terms of original variables

被引：9

作者：

Daszykowski, M

Stanimirova, I

Walczak, B

Coomans, D

机构：

[1] Silesian Univ, Inst Chem, Dept Chemometr, PL-40006 Katowice, Poland

[2] Vrije Univ Brussels, FABI, ChemoAC, B-1090 Brussels, Belgium

[3] James Cook Univ N Queensland, Sch Math & Phys Sci, Stat & Intelligent Data Anal Grp, Townsville, Qld 4814, Australia

来源：

CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS | 2005年 / 78卷 / 1-2期

关键词：

exploratory analysis; CART; MRT; projection pursuit; projection index; kurtosis;

D O I：

10.1016/j.chemolab.2004.12.002

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

This manuscript shows the usefulness of Projection Pursuit (PP) and Multivariate Regression Trees (MRT) for analytical data exploration. Additionally, features of Projection Pursuit and kurtosis as a projection index are presented. The ability of Projection Pursuit to discover groups in the data is compared to classical Principal Component Analysis (PCA). Moreover, it is also demonstrated how the presence of groups in the data can be explained in terms of explanatory variables with the aid of Projection Pursuit and Multivariate Regression Trees. Neither Projection Pursuit nor Multivariate Regression Trees are commonly used for exploring chemical data however, they are able to enrich to a high extent the interpretation. (c) 2004 Elsevier B.V. All fights reserved.

引用

页码：19 / 29

页数：11

共 28 条

[1]

Breiman L., 1998, CLASSIFICATION REGRE

[2]

Croux C., 1996, COMPSTAT. Proceedings in Computational Statistics. 12th Symposium, P211

[3] A journey into low-dimensional spaces with autoassociative neural networks [J].