Detection of inhomogeneities in sets of NIR spectra

被引：40

作者：

Centner, V

Massart, DL

deNoord, OE

机构：

[1] FREE UNIV BRUSSELS,CHEMOAC,B-1090 BRUSSELS,BELGIUM

[2] KONINKLIJKE SHELL EXPTL PROD LAB,SHELL RES BV,NL-1030 BN AMSTERDAM,NETHERLANDS

来源：

ANALYTICA CHIMICA ACTA | 1996年 / 330卷 / 01期

关键词：

multivariate calibration; chemometrics; near-infrared spectrometry; inhomogeneity;

D O I：

10.1016/0003-2670(96)00150-X

中图分类号：

O65 [分析化学];

学科分类号：

070302 ; 081704 ;

摘要：

Methods for the detection of outliers and clusters in data sets for multivariate calibration are discussed in this study. Replicate as well as sample outliers have been investigated. The Cochran test applied to the sum of absorbances is shown to identify outlying replicates. Sample outliers were detected on latent variables (PCs) or by using Rao's statistic combined with Grubbs' and Dixon's tests. The conclusions were then verified by the modelling. The tendency to cluster was evaluated by Hopkins statistic.

引用

页码：1 / 17

页数：17

共 14 条

[1]

[Anonymous], 1989, MULTIVARIATE CALIBRA

[2] THE DISTRIBUTION OF THE RATIO, IN A SINGLE NORMAL SAMPLE, OF RANGE TO STANDARD DEVIATION [J].