[R-sig-ME] How to extract num. of observations and groups used in a mixed model using gee package

Ben Bolker bbolker at gmail.com
Tue Apr 24 17:01:24 CEST 2012


On 12-04-24 10:54 AM, Robert Kushler wrote:
> 
> The following lines from the gee help page ...
> 
> Details
> 
> Though input data need not be sorted by the variable named "id", the
> program will interpret physically contiguous records possessing the same
> value of id as members of the same cluster. Thus it is possible to use
> the following vector as an id vector to discriminate 4 clusters of size
> 4: c(0,0,0,0,1,1,1,1,0,0,0,0,1,1,1,1).
> 
> 
>  ... indicate that the proposed strategy may not yield a correct
> answer.  In fact, the above suggests
> that you could get different results depending on whether or not your
> data set is sorted.  I find
> this quite disturbing, but haven't taken the time to test it.
> 
> Regards,   Rob Kushler

  Yow.  Good catch.

 Maybe length(rle(geemodel$id)$length) would work?


> (id <- rep(c(0,1,0),each=3))
[1] 0 0 0 1 1 1 0 0 0
> length(rle(id)$lengths)
[1] 3
> (id <- rep(c(0,1,0,1),each=3))
 [1] 0 0 0 1 1 1 0 0 0 1 1 1
> length(rle(id)$lengths)
[1] 4

> 
> 
> 
> 
> 
> On 4/24/2012 10:25 AM, David Winsemius wrote:
>>
>> On Apr 24, 2012, at 9:46 AM, Ben Bolker wrote:
>>
>>> david oseguera montiel <oseguera.david at ...> writes:
>>>
>>>>
>>>> I meant the grouping factor in the model indicated by id. From gee
>>>> documentation 'id = a vector which identifies the clusters..'. So in
>>>> the
>>>> warpbreaks data set, id = wool has 54 obs, in two clusters (groups).
>>>
>>> [snip]
>>>
>>>>>> David,
>>>>>> Thank you very much. Number of observations solved, but I am still
>>>>>> missing how to find out the number of groups used. In the warpbreaks
>>>>>> data set these are 2. Any ideas. Thank you.
>>>>>>
>>>>> I must be unclear what you mean by "number of groups".
>>>
>>>
>>> Didn't Chuck Cleland's answer work, i.e. length(unique(model$id)) ... ?
>>
>> It didn't work for me.
>> length(unique(summgee$id))
>> [1] 0
>>  > summgee$id
>> NULL
>>
>> See my other reply for what did work. I suppose it could be that both
>> the OP and I are behind on gee versions. Mine is
>> 4.13-17 and I see that there is a more recent binary on CRAN. With a
>> (slightly out-of-date) version loaded, I am loath
>> to detach because getting things to all sync up is a bit of a hassle
>> and error-prone without a fresh start and I have
>> other work time-consuming running in this session.
>>



More information about the R-sig-mixed-models mailing list