On 18/03/2015 15:38, peter dalgaard wrote:

On 18 Mar 2015, at 15:11 , Annalisa Orenti <annalisaore...@libero.it> wrote:

Dear R-Users,
I need to fit a regression by means of S-estimator, but I found an
inconsistency in lqs function in MASS package.
When recalling coefficient estimates two discordant answers are obtained by
coef() function and by $coef object:
As an example I report here the results of fitting regression S-estimator on
stackloss data:

library(MASS)
Sestim<-lqs(stack.loss ~ Air.Flow + Water.Temp + Acid.Conc., data = stackloss,
method = "S")
coef(Sestim)
(Intercept)            Air.Flow   Water.Temp   Acid.Conc.
-35.37610619   0.82522124   0.44247788  -0.07964602
Sestim$coef
                           Air.Flow   Water.Temp   Acid.Conc.
-36.92530315   0.84957479   0.43047554  -0.07354065

Do you have any suggestions?


The immediate reason is that Sestim contains both $coef and $coefficients and 
coef() extracts the latter.

Why this is so escapes me. It looks like it might be a blunder, but the package 
author isn't usually prone to blundering...

It is a side-effect of changes in R since the code was written (for S). Use Sestim$coef until MASS is updated (which will be before 3.2.0 is released).

-pd

Thank you for your help.
Best regards.
Annalisa

______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.



--
Brian D. Ripley,                  rip...@stats.ox.ac.uk
Emeritus Professor of Applied Statistics, University of Oxford
1 South Parks Road, Oxford OX1 3TG, UK

______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Reply via email to