[R-sig-eco] Cluster Analysis

[Jari Oksanen]

On 19/05/11 20:20 PM, "Francisco Cabrera Rodriguez" <fffcccrrr at yahoo.com>
wrote:

> I would like
> to perform a cluster analysis to data of presence absence.
> Which distance
> and group linkage methods are best?

Francisco,

Nobody knows. Honestly. Some people may have opinions. Even strong opinions.
Perhaps they will help you.

We can start from the easier part: what makes a good class, in your opinion?
Should they be uniform? or compact? or connected? or distinct? These
criteria are contradictory and you must make your pick.

Reading a good book may help.

Cheers, Jari Oksanen