[R] plot hclust object

[Greg Snow]

I don't know of any tools that automate this process.  For small
sample sizes it may be easiest to just do this by hand, for large
sample sizes that plot will probably be to complicated to make sense
of.  There may be a range of moderate sample sizes for which
automation (or partial automation) would be helpful.  The hclust
object has a component of "height" which is an indicator of the
distance between 2 components being combined into a cluster, you could
convert this into a distance matrix (or extract the distance matrix
used to do the clustering if it is available) and then use
multidimensional scaling (cmdscale function is one option) to produce
a 2 dimensional set of points.  Drawing the circles/ellipses/ovals
will be more difficult, possibly generate a cloud of normal points, or
a small circle, around each point with the variability/radius low
enough that the clouds are unlikely to overlap, then find the convex
hull (chull function) for the points within a cluster and draw that
(it will be a polygon rather than a smooth curve).  The gBuffer
command in the rgeos package may be another way to create polygons
around the points in a group.

On Mon, Oct 27, 2014 at 5:42 AM, David Feitosa <davidfeitosa at gmail.com> wrote:
> Hello!
>
> I have a code that creates an hclust object.
> After the object creation I plot the object as a dendrogram,
> similar to the left image of this link:
>
> http://www.cs.jhu.edu/~razvanm/fs-expedition/hclust-example.png
>
> I would like to create another image, but similar to the right,
> as a set of nested  dots and elipses/circles.
>
> Anybody knows how to do this?
>
> Thanks in advance.
>
> David Feitosa
>
> (\_(\
> (=°;°)
> (("")("")
>
>         [[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.



-- 
Gregory (Greg) L. Snow Ph.D.
538280 at gmail.com