[R] Using function 'boot on a bi-polar sample'

[Prof Brian Ripley]

On Tue, 10 May 2005, Briggs, Meredith M wrote:

> I'm not sure I'm using boot correctly:

You are not using argument `i'.  Have you read the reference: `boot' is
support software for a book?  (Or even the definition of `statistic' on 
the help page?)  The bootstrap sample in your case is x[i], not x.

> I have a list of values for a variable in BUCKET[, j] I want to use 
> function 'boot' to estimate a confidence interval on the mean of the 
> non-zero data. The data can be bi-polar or skewed. Is this the correct 
> use of boot to establish a mean and standard deviation or median and 
> percentiles?
>
>
> 	x<-BUCKET[, j]
>          	mean.fun<-function(x,i) {mean(x[x>0])}
> 	median.fun<-function(x,i) {median(x[x>0])}
> 	print(mean.fun(x))
> 	print(median.fun(x))
>      	B.mean<-boot(x,mean.fun,R=999)
> 	B.median<-boot(x,mean.fun,R=999)
>      	print(B.mean)
> 	print(B.median)
>
> I need the 'boot' function to sample repeatedly from the density 
> distribution of the raw data and calculate means, medians etc.
>
> How do I pull out the mean, median, deviations and percentiles into 
> variables within R?


-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
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