[R-sig-eco] Problem with result PCA

[THIOULOUSE JEAN]

Le 27 août 2011 à 11:50, momadou sow a écrit :
> Hi, 
> I created a PCA (see attached file). I use this code: 
>> library(FactoMineR)
>> res.pca<-PCA(acp)
>  I use satellite indexes and two indexes of vegetation water content (FMC and EWT) measured in field. For the satellite indexes some values are negative while all the values for FMC and EWT are positive. 
> Now when I run my code, I see FMC and EWT are in the negative part in the PCA between [0,-1] (please see attached file).
> Is it normal to have variables with positive values in the negative part of PCA while the variables with negative values are in the positive part?
> Thank you in advance. 

The direction of PCA axes is arbitrary. This means you could very well have obtained variables with positive values in the positive part of PCA. This can change according to the software and even to the computer model that you use for computations. Of course, the sign of item scores also change, so the resulting interpretations are unchanged.

Jean