[R] Problems with boot package (empinf returns NA)

[Roger Peng]

How many bootstrap resamples did you use in your original call to `boot'?
I sometimes get this error when the number of resamples is small.  Try
increasing the value for the `R' parameter in `boot' and rerun the bca.ci
function.  For example,

> set.seed(100)
> f <- function(x, i) median(x[i])
> bca.ci(boot(x, f, R = 10))
Error in if (!all(rk > 1 & rk < R)) warning("Extreme Order Statistics used
as Endpoints") : 
        missing value where TRUE/FALSE needed

But the following works,

> set.seed(100)
> bca.ci(boot(x, f, R = 1000))
     conf                                   
[1,] 0.95 20.15 970.16 -0.3172107 0.06934598


-roger
_______________________________
UCLA Department of Statistics
rpeng at stat.ucla.edu
http://www.stat.ucla.edu/~rpeng

On 31 Jan 2003, Ernesto Jardim wrote:

> Hi
> 
> I'm using boot package for some analysis on linear regression
> coeficients. My problem is that I can not compute bca intervals, I get
> an error message
> 
> > bca.ci(blm8901,index=1)
> Error in if (!all(rk > 1 & rk < R)) warning("Extreme Order Statistics
> used as Endpoints") : 
>         missing value where logical needed
> 
> The problem is the empinf.reg function that is returning Na when
> performing a glm regression, statement
> 
> beta <- coefficients(glm(t ~ X))[-1]
>   
> I don't see anything strange in t or X, but I'm missing something for
> sure. 
> 
> Can someone give me an hint on this ?
> 
> Thanks
> 
> EJ
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> http://www.stat.math.ethz.ch/mailman/listinfo/r-help
>