R help - Apr 2006 - Bootstrap error message: Error in statistic(data, origina l, ...) : unused argument(s) ( ...) [Broadcast]

I quoted the relevant part of the documentation for you.  Have you actually
try to read what it says?
 
Sure, you don't get any error, but have you checked whether any
bootstrapping was actually done?  Most of those functions are generics, thus
having the "..." argument that can take anything.  Doesn't mean
they will be
used.  See if the following helps:
 
> x <- 1:10
> myMean <- function(x, idx) mean(x[idx])
> library(boot)
> x.boot <- boot(x, mean, R=10)
> str(x.boot)
List of 11
 $ t0       : num 5.5
 $ t        : num [1:10, 1] 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5
 $ R        : num 10
 $ data     : int [1:10] 1 2 3 4 5 6 7 8 9 10
 $ seed     : int [1:626] 403 101 106329237 1120199471 -647917002 -657447328
355480739 291889089 -398656592 -2023350578 ...
 $ statistic:function (x, ...)  
 $ sim      : chr "ordinary"
 $ call     : language boot(data = x, statistic = mean, R = 10)
 $ stype    : chr "i"
 $ strata   : num [1:10] 1 1 1 1 1 1 1 1 1 1
 $ weights  : num [1:10] 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
 - attr(*, "class")= chr "boot"
> x.boot2 <- boot(x, myMean, R=10)
> str(x.boot2)
List of 11
 $ t0       : num 5.5
 $ t        : num [1:10, 1] 5.7 6.4 5.4 4.9 7.2 4.6 4.4 5.4 5.7 5.9
 $ R        : num 10
 $ data     : int [1:10] 1 2 3 4 5 6 7 8 9 10
 $ seed     : int [1:626] 403 201 106329237 1120199471 -647917002 -657447328
355480739 291889089 -398656592 -2023350578 ...
 $ statistic:function (x, idx)  
  ..- attr(*, "source")= chr "function(x, idx) mean(x[idx])"
 $ sim      : chr "ordinary"
 $ call     : language boot(data = x, statistic = myMean, R = 10)
 $ stype    : chr "i"
 $ strata   : num [1:10] 1 1 1 1 1 1 1 1 1 1
 $ weights  : num [1:10] 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
 - attr(*, "class")= chr "boot"
 
Andy

-----Original Message-----
From: Michael [mailto:comtech.usa@gmail.com] 
Sent: Thursday, April 20, 2006 12:19 PM
To: Liaw, Andy
Cc: R-help@stat.math.ethz.ch
Subject: Re: [R] Bootstrap error message: Error in statistic(data, origina
l, ...) : unused argument(s) ( ...) [Broadcast]


btw, if you change "myFun" to any R internal function, such as
"cov", or
"corr", it can run successfully...


On 4/20/06, Michael <comtech.usa@gmail.com
<mailto:comtech.usa@gmail.com> >
wrote: 

Andy,

I've noticed there should be a "weight" or "frequency"
somewhere... but my
function does not need it.

I have tried to cheat it by declareing a weight, but not using it:
> myFun=function(X, w) 
> {
>      return(mean(X[, 1])/var(X[, 2]));
> }
>
> bootResults=boot(X, myFun, R=10000, stype='w'); 
The result is zero:

Bootstrap Statistics :
    original  bias    std. error 
t1* 2.305412       0           0



On 4/20/06, Liaw, Andy < andy_liaw@merck.com
<mailto:andy_liaw@merck.com> >
wrote: 
> -----Original Message-----
> From: r-help-bounces@stat.math.ethz.ch
<mailto:r-help-bounces@stat.math.ethz.ch> 
> [mailto: r-help-bounces@stat.math.ethz.ch
<mailto:r-help-bounces@stat.math.ethz.ch> ] On Behalf Of
Michael
> Sent: Thursday, April 20, 2006 3:50 AM
> To: R-help@stat.math.ethz.ch  <mailto:R-help@stat.math.ethz.ch> 
> Subject: [R] Bootstrap error message: Error in
> statistic(data, original, ...) : unused argument(s) ( ...) [Broadcast] 
>
>
> Dear colleagues,
>
> I've been swamped and fighting with error for a few hours but
> still desperately having absolutely no clue:
You really don't have to do that, but just RTFM instead.  ?boot says: 

statistic   A function which when applied to data returns a vector
            containing the statistic(s) of interest. [...]
            In all other cases statistic must take at least two
            arguments. The first argument passed will always be 
            the original data. The second will be a vector of
            indices, frequencies or weights which define the
            bootstrap sample. [...]

Your "myFun" clearly does not fit that description.  boot() tried to
call
"myFun" with a second argument, but "myFun" doesn't know
what to do with a
second argument, and that's where you get the error.

Andy

> What's wrong with my bootstraping code? 
>
> Thanks a lot!
>
> ------------------------
> Error Message:
>
> > bootResults=boot(X, myFun, R=10000);
> Error in statistic(data, original, ...) : unused argument(s) ( ...) 
>
> ------------------------
> My code is:
>
> X=cbind(column_vector_1, column_vector_2);
> myFun=function(X)
> {
>      return(mean(X[, 1])/var(X[, 2]));
> }
> 
> bootResults=boot(X, myFun, R=10000);
>
> ---------------------
>
>       [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help@stat.math.ethz.ch <mailto:R-help@stat.math.ethz.ch>  mailing
list
> https://stat.ethz.ch/mailman/listinfo/r-help
<https://stat.ethz.ch/mailman/listinfo/r-help> 
> PLEASE do read the posting guide!
> http://www.R-project.org/posting-guide.html
<http://www.R-project.org/posting-guide.html> 
>
>

----------------------------------------------------------------------------
--
Notice:  This e-mail message, together with any attachments,...{{dropped}}

Hey everybody. 

I found the solution to the problem - at least for univariate statistics. In
order to use the bootstrap method for an arbitrary statistic, the syntax for
the function has to be in the form:

Given a univariate formula with argument 'x':

myfunction <- function(x,p){formula(x[p])}

The idea of the boot() function seems to be to generate a randon vector of
integers (representing the bootstrapped samples) and evaluate the data
matrix at said vector, evaluate the function, store the bootstrapped
statistic, and repeat. That's why it needs the second argument, and also why
it gives the same value of t* when no such argument is present (i.e. "all
values are equal to 21.10958105")

Cheers!
-- 
View this message in context:
http://r.789695.n4.nabble.com/Re-Bootstrap-error-message-Error-in-statistic-data-origina-l-unused-argument-s-Broadcast-tp798359p3043991.html
Sent from the R help mailing list archive at Nabble.com.