Local updating is what I would have recommended.
But predict() will work correctly with NA values in the fit if the latter
was done with na.action=na.exclude. I would find it clearer to use
fitted() here, which also works for na.action=na.exclude.
On Wed, 24 Oct 2007, Christos Hatzis wrote:
> Another way to do this without messing with environments is to update the
> data and formula locally within the function and re-run the regression on
> the updated model/data:
>
> tukey.test <- function(m) {
> ud <- data.frame( m$model, pred2=m$fitted.values^2 )
> uf <- update.formula(formula(m$terms), ~ . + pred2)
> summary(lm(uf, ud))$coef
> }
>
>> data(BOD)
>> m1 <- lm(demand~Time,BOD)
>> tukey.test(m1)
> Estimate Std. Error t value Pr(>|t|)
> (Intercept) 12.5854086 5.0344850 2.4998403 0.0877190
> Time 7.5390634 6.1271204 1.2304415 0.3062126
> pred2 -0.1096744 0.1148654 -0.9548081 0.4101142
>
>
>> -----Original Message-----
>> From: [EMAIL PROTECTED]
>> [mailto:[EMAIL PROTECTED] On Behalf Of Sandy Weisberg
>> Sent: Wednesday, October 24, 2007 3:44 PM
>> To: [EMAIL PROTECTED]
>> Subject: [R] scoping problem
>>
>> I would like to write a function that computes Tukey's 1 df
>> for nonadditivity. Here is a simplified version of the
>> function I'd like to
>> write: (m is an object created by lm):
>>
>> tukey.test <- function(m) {
>> m1 <- update(m, ~.+I(predict(m)^2))
>> summary(m1)$coef
>> }
>>
>> The t-test for the added variable is Tukey's test. This won't work:
>>
>> data(BOD)
>> m1 <- lm(demand~Time,BOD)
>> tukey.test(m1)
>>
>> Error in predict(m) : object "m" not found
>>
>> This function doesn't work for two reasons:
>> 1. The statement m1 <- update(m, ~.+I(predict(m)^2))
>> can't see 'm' in the call to predict.
>> 2. If in creating m missing values had been present,
>> then predict(m), even if it could be computed, could be of
>> the wrong length.
>>
>> Can anyone help?
>>
>>
>>
>> --
>> Sanford Weisberg, [EMAIL PROTECTED]
>> Office and mailing address:
>> University of Minnesota, School of Statistics
>> 312 Ford Hall, 224 Church St. SE, Minneapolis, MN 55455
>> 612-625-8355, FAX 612-624-8868
>>
>> St. Paul office:
>> 146 Classroom-Office Building, 612-625-8777
>>
>> ______________________________________________
>> R-help@r-project.org mailing list
>> 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.
>>
>>
>
> ______________________________________________
> R-help@r-project.org mailing list
> 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, [EMAIL PROTECTED]
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595
______________________________________________
R-help@r-project.org mailing list
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.
