[R] how to change the "quantile" method in bwplot

[Deepayan Sarkar]

On Tue, Jul 21, 2009 at 7:47 AM, Jun Shen<jun.shen.ut at gmail.com> wrote:
> Uwe,
>
> Thank you for your reply.  I am still not very clear about the meanings of
> the arguments in the stats function.   To make it clearer, quantile() uses
> type=7 as default method. I believe this is the method bwplot() uses to
> calculate the quantiles. I want to use type=6 method for bwplot(). How do I
> achieve that? Thanks again.

Maybe this will be clearer: bwplot() uses the boxplot.stats() function
to compute the "quantiles" used, which in turn uses fivenum(), which
has its own quantile calculation (and does not explicitly use
quantile()). There is no easy way to allow for type=6 etc. here.

bwplot() allows you to replace boxplot.stats() and provide your own
alternative. So what you need to do is:

(1) write a function, say, 'my.boxpot.stats', that takes the same
arguments as boxplot.stats() and returns a similar result, but using
your preferred calculation for the quantiles. There are many ways to
do this.

(2) plug in this function into the bwplot() call; e.g. bwplot(...,
stats = my.boxplot.stats)

-Deepayan


>
> Jun
>
> 2009/7/21 Uwe Ligges <ligges at statistik.tu-dortmund.de>
>
>>
>>
>> Jun Shen wrote:
>>
>>> Hi, everyone,
>>>
>>> Since quantile calculation has nine different methods in R, I wonder how I
>>> specify a method when calling the bwplot() in lattice. I couldn't find any
>>> information in the documentation. Thanks.
>>>
>>>
>>
>> bwplot() uses the panel function panel.bwplot() which allows to specify a
>> function that calculates the statistics in its argument stats that defaults
>> to boxplot.stats(). Hence you can change that function.
>>
>> Example with some fixed values:
>>
>> bwplot( ~ 1:10,
>>    stats = function(x, ...)
>>        return(list(stats=1:5, n=10, conf=1, 10, out=integer(0)))
>> )
>>
>>
>> Uwe Ligges
>>
>
>        [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at 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.
>