Re: PLS MorphoJ TPSPLS

[morphmet_moderator]

----- Forwarded message from "F. James Rohlf" -----

Date: Wed, 9 Jan 2013 19:24:26 -0500
From: "F. James Rohlf"
Reply-To: "F. James Rohlf"
Subject: Re: PLS MorphoJ TPSPLS
To: morphmet@morphometrics.org

If the variables are in a mixture of units then you should standardize. 

-------------
F. James Rohlf, E&E Stony Brook University
Sent from my mobile device

-------- Original message --------
From: morphmet_modera...@morphometrics.org
Date:
To: morphmet@morphometrics.org
Subject: Re: PLS MorphoJ TPSPLS



----- Forwarded message from carlo.mel...@unina.it -----

     Date: Wed, 9 Jan 2013 04:16:49 -0500
      From: carlo.mel...@unina.it
      Reply-To: carlo.mel...@unina.it
      Subject: Re: PLS MorphoJ TPSPLS
      To: morphmet@morphometrics.org

Dear Marisol,

in tpsPLS under the section "Option" you should see if it is ticked 
the voice "Standardize variables" or "Standardize Partial 
Warps".....to my understanding MorphoJ does not standardize variables 
while tpsPLS does this automatically.

I would like to take advantage of this email also to ask:

when doing PLS of variables vs shape, is it more correct to 
standardize the variables or not to?
I find out that when doing PLS between shape and environmental 
variables standardisation changes considerably the Singular Warp 
loadings.

Cheers

Carlo

morphmet_modera...@morphometrics.org ha scritto:

>
> ----- Forwarded message from marisol anzelmo  -----
>
>      Date: Wed, 2 Jan 2013 19:44:17 -0500
>       From: marisol anzelmo
>       Reply-To: marisol anzelmo
>       Subject: FW: PLS MorphoJ TPSPLS
>       To: morphmet@morphometrics.org
>
>  Hi,
>
> I collected landmarks and semilandmarks on two curves. I slide 
> semilandmarks in TPSRelw. I want to do Partial Least Square between 
> the two curves. I performed PLS in two softwares: MorphoJ and 
> TPSPLS. I detected that the adjusted coordinates with both softwares 
> make the PLS are very similar. However, the final results I get are 
> completely different . In TPSPLS I have a PLS 1 with a correlation 
> of 0.4  (including or not including the uniform component) while 
> MorphoJ PLS 1 is close to 0.9. I wonder why I get such different 
> results if coordenates adjusted on both programs are doing the 
> analysis are very similar.
>
> Regards,
>
> Marisol
>
> ----- End forwarded message -----
>
>

----- End forwarded message -----

----- End forwarded message -----