[R] hierarhical cluster analysis of groups of vectors

[Prof Brian Ripley]

On Tue, 29 May 2007, Anders Malmendal wrote:

> Thanks.
> The vectors are produced by PLS-discriminant analysis between groups and
> the vectors within a group are simply different measurements of the same
> thing. What I need is a measure of how the different groups cluster
> (relative to each other). (I assume that I can do some averaging after
> applying dist, but I can not  find the information on how to do it.)

I don't think anyone can tell you that: it is a matter of judgement. 
What you need is a dissimilarity on your groups.

Assuming your vectors are numeric (you didn't say) you could use 
Mahalanobis distance between the centroids, with within-group covariance 
as the variance matrix.  Often that works well, but not always, and you 
might prefer Euclidean distance between centroids, or minimum Euclidean or 
Mahalanobis distance ....


> Best regards
> Anders
>
>
> Rafael Duarte wrote:
>> It seems that you have already groups defined.
>> Discriminant analysis would probably be more appropriate for what you
>> want.
>> Best regards,
>> Rafael Duarte
>>
>>
>>
>> Anders Malmendal wrote:
>>
>>> I want to do hierarchical cluster analysis to compare 10 groups of
>>> vectors with five vectors in each group (i.e. I want to make a
>>> dendogram showing the clustering of the different groups). I've
>>> looked into using dist and hclust, but cannot see how to compare the
>>> different groups instead of the individual vectors. I am thankful for
>>> any help.
>>> Anders
>>>
>>> ______________________________________________
>>> R-help at stat.math.ethz.ch 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.
>>>
>>>
>>
>>
>
>
>

-- 
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