Petr Pikal replied:
[...]
> The following works
> >
> > results <- boot(data=test, statistic=bs, R=1000, A~B+C+D+C*D)
> >
>
> Actually it does not work either
>
Correct, but I _did_ get it to work shortly before my initial post (sorry
for not showing it, but I didn't save the output - silly me).
What seems to happen is that I successfully run the version that _does_
work, which I then follow up with successful calls to
-plot(results,index=#)- and -boot.ci(results,type="bca",index=#)-, then I
follow up with the unsuccessful version of the code, which won't run and
produces the error, but when I go back to run the original code, this too
produces the same error. This, to me at least, is very weird behaviour for
R.
>
> > results <- boot(data=test, statistic=bs, R=1000, A~B+C+D+C*D)
> Error in data[indices, ] : incorrect number of dimensions
> >
>
> I am not sure but I suspect your bs function expects some indices vector
> and it is somehow not in accordance with your data.
>
This code, using different simulated data _does_ run, however:
> mytest<-data.frame(Y=rnorm(500, mean=60, sd=1), X1=sample(c("AAA", "BBB",
> "CCC", "DDD", "None", NA), size=500, replace=TRUE), X2=sample(c(0.88, 0.99,
> 1.43, 1.22, 1.1), size=500, replace=TRUE), X3=sample(c("Yes", "No"),
> size=500, prob=c(0.6,0.4), replace=TRUE))
> bs <- function(formula, data, indices) {+ mytest <- data[indices,]+
> fit <- lm(formula, data=mytest)+ return(coef(fit)) + }> results <-
> boot(data=mytest, statistic=bs, R=1000, formula=Y~X1+X2+X3)> results
ORDINARY NONPARAMETRIC BOOTSTRAP
Call:
boot(data = mytest, statistic = bs, R = 1000, formula = Y ~ X1 +
X2 + X3)
Bootstrap Statistics :
original bias std. error
t1* 60.10691271 -0.0012790917 0.28940597
t2* -0.01959846 -0.0033038861 0.14376648
t3* 0.24000669 -0.0016111232 0.16163714
t4* 0.19158141 -0.0015946959 0.14849004
t5* 0.08128826 -0.0013207129 0.14309419
t6* -0.25319061 0.0002987212 0.22874411
t7* 0.03443774 0.0047553487 0.09620542
How do you explain that? Or, indeed, this, which takes two variables with
simulated proportions:
> mytest<-data.frame(Y=rnorm(500, mean=60, sd=1), X1=sample(c("AAA", "BBB",
> "CCC", "DDD", "None", NA), size=500, replace=TRUE), X2=sample(c(0.88, 0.99,
> 1.43, 1.22, 1.1), size=500, replace=TRUE), X3=sample(c("Yes", "No"),
> size=500, prob=c(0.6,0.4), replace=TRUE), X4=sample(c(0,1), size=500,
> prob=c(0.25,0.75), replace=TRUE))> bs <- function(formula, data, indices) {+
> mytest <- data[indices,]+ fit <- lm(formula, data=mytest)+
> return(coef(fit)) + }> results <- boot(data=mytest, statistic=bs, R=1000,
> formula=Y~X1+X2+X3+X4)> results
ORDINARY NONPARAMETRIC BOOTSTRAP
Call:
boot(data = mytest, statistic = bs, R = 1000, formula = Y ~ X1 +
X2 + X3 + X4)
Bootstrap Statistics :
original bias std. error
t1* 60.28192671 -0.015122589 0.3016916
t2* -0.12397654 -0.005209852 0.1558575
t3* -0.15597837 -0.007948155 0.1593787
t4* -0.13420834 -0.004417036 0.1499142
t5* -0.03049650 -0.003097340 0.1600422
t6* -0.07926231 0.012471978 0.2477136
t7* -0.10564206 -0.002528748 0.1028371
t8* 0.01627027 0.005540313 0.1141790
And yet it won't run with my simulated dataset, despite the fact that it's
quite similar to this? Sorry, I'm not buying it.
Answers please!
Clive Nicholas (clivenicholas.posterous.com
[Please DO NOT mail me personally here, but at <clivenicho...@hotmail.com>.
Please respond to contributions I make in a list thread here. Thanks!]
"My colleagues in the social sciences talk a great deal about methodology.
I prefer to call it style." -- Freeman J. Dyson
[[alternative HTML version deleted]]
______________________________________________
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.
