Hi all.
I am seeking for a statistical suggestion.
My data set comprises 382 measures each having 169 variables. Each measure
is the outcome of a nuclear magnetic resonance experiment, so each of the
169 points has the same unit.
As I want to do some multivariate calibration using these data, I checked
whether some multivariate outliers existed. I calculated Mahalanobis
distances and did a
qqplot(qchisq(ppoints(382),169),XMaha)
where XMaha contained my Mahalanobis distances. The points on my QQ plot
were nicely aligned onto a y=x line, so no outliers.
For some reasons I need to scale my data set so as that each measure
has the same initial point, i.e. the first point of each measure must be
1. To be clear I did:
x1<-x[,1:169]/x[,1]
where x and x1 are the unscaled and scaled sata sets, respectively.
However, after doing this (and eliminating the first point of each
measure), the points in the new QQ plot -
qqplot(qchisq(ppoints(382),168),XMahaNew) -
were still aligned, although they were no more along the y=x line.
Am I doing something statistically wrong ?
Thank you all in advance.
Mauro
===========================================Mauro Andrea Cremonini, PhD
Department of Food Science
University of Bologna
Via Ravennate 1020 - 47023 - Cesena - Italy
FAX:+39.0547.382348 - Phone:+39.0547.636106
e-mail: mac at foodsci.unibo.it
http://www.foodsci.unibo.it/cremonini
===========================================
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
