At 11:40 13/11/2001 +0100, Agustin Lobo wrote:
> > Non, on ne trouve pas cette méthode dans ADE-4 qui ne contient que des
> > méthodes basées sur la géométrie euclidienne.
>
>May I ask why? Why do we go on pretending things are linear?
>I have the impression that this type of non-linear
>analysis is not often used, it's just a matter of
>complexity in the programming and/or computational requirements
>or are there other (theoretical) drawbacks?
>
>In the paper you mention (in the 70's), it says
>"The distortion in indirect gradient analysis is a consequence of
>assumptions in conflict with knowledge of species distributions"
La question est importante
Pour moi, une méthode linéaire est une méthode basée sur l'algèbre linéaire
Les choses ne sont pas linéaire, l'algèbre peut l'être !
"linear" is "linear algebra"
Il faut voir le très important article :
Ter Braak, C. J. F. 1985. Correspondence analysis of incidence and
abundance data : properties in terms of a unimodal reponse model.
Biometrics 41:859-873.
"In this paper correspondence analysis is shown to approximate the maximum
likelihood solution of explicit unimodal response models in one latent
variable"
Etonnant, non ?
This archive was generated by hypermail 2b30 : Thu Jul 04 2002 - 09:39:14 MEST