[R] lattice/ylim: how to fix ylim[1], but have ylim[2] dynamically calculated?

[David Winsemius]

On Feb 14, 2010, at 3:40 PM, Johannes Graumann wrote:

> David Winsemius wrote:
>
>>
>> On Feb 14, 2010, at 10:33 AM, Johannes Graumann wrote:
>>
>>> Hello,
>>>
>>> When drawing "barcharts", I find it not helpful if ylim[1] != 0 -
>>> bars for a
>>> quantity of 0, that do not show a length of 0 are quite non- 
>>> intuitive.
>>>
>>> I have tried to study
>>>> library(lattice)
>>>> panel.barchart
>>> but am unable to figure out where ylim is taken care of and how one
>>> might
>>> fix ylim[1] to 0 for barcharts ...
>>>
>>> Can anyone point out how to tackle this?
>>
>> Looking at Sarkar's "Lattice" text in chapter 8 section 3 "Limits and
>> Aspect Ratio", it appears from subsection 1 that the prepanel  
>> function
>> can used to supply values of xlim and ylim values. From subsection 2
>> he clarifies that xlim and ylim can also be specified on a per panel
>> basis (and here I am guessing that this would be within a scales
>> argument) when relation="free". At the end of that section he offers
>> two examples using ylim: the first is not plotted but the second uses
>> the prepanel mechanism for Fig 8.1 and that is probably available on
>> the Lattice website.
>>
>> In the same subsection is offered an alternative to specifying an
>> explicit scales$y$limits to be interpreted as ylim values.
>>
>> My hope it that these ideas and references will be of some use in
>> identifying productive places to look for further documentation.
>
> I am quite certain that this is the most elaborately worded version of
> "RTFM" I have ever come across.

Well, not really, it was a version of ... here's my best untested wild  
guess from someone who persistently struggles with trying to get panel  
arguments right.

> I shall go and do so.
>
> Sincerely, Joh
-- 

David Winsemius, MD
Heritage Laboratories
West Hartford, CT