[GRASS-user] Comparison between "i.pca" and R's "prcomp()": explanations and questions

[Nikos Alexandris]

On Fri, 2009-02-06 at 14:12 +0100, Nikos Alexandris wrote:
> Eigen values:
> ( -0.48 -0.45 -0.28 -0.45 -0.45 -0.29 )
> ( -0.48 0.24 0.37 -0.53 0.26 0.47 )
> ( -0.52 -0.37 -0.28 0.50 0.49 0.16 )
> ( 0.45 -0.32 -0.50 -0.33 0.10 0.58 )
> ( 0.03 -0.48 0.54 0.29 -0.45 0.42 )
> ( 0.26 -0.52 0.40 -0.28 0.52 -0.39 )

>        b2.2006 b6.2006 b7.2007 b2.2007 b6.2007 b7.2007
> PC1: (  -0.48   -0.45   -0.28   -0.45   -0.45   -0.29 )
> PC2: (  -0.48    0.24    0.37   -0.53    0.26    0.47 )
> PC3: (  -0.52   -0.37   -0.28    0.50    0.49    0.16 )
> PC4:(    0.45   -0.32   -0.50   -0.33    0.10    0.58 )
> PC5: (   0.03   -0.48    0.54    0.29   -0.45    0.42 )
> PC6: (   0.26   -0.52    0.40   -0.28    0.52   -0.39 )
> 
> 
> # The signs of the loadings are *probably* arbitrary. As one can see,
> the i.pca-produced loadings differ from R's output (in fact they are
> sometimes, on a per-column basis, exactly the opposite).

Well, not in this example :-). But they *really* do differ.