[GRASS-user] Signature not invertible

[Markus Neteler]

On Thu, Apr 15, 2010 at 11:50 AM, joel dinis <dinis.joel at gmail.com> wrote:
> Hello list,
>
> I'm with a problem running the i.gensig.
> I do:
>
> i.gensig trainingmap=areasraster at Portugal_9 group=grupo at Portugal_9
> subgroup=subgrupo signaturefile=assinaturascomareas
>
> where areasrastes has 4 classes.
> The *r.stats -c input=areasraster at Portugal_9* is
> 1 92
> 2 234
> 3 2434
> 4 3016
> * 287024
>
> And I get the following:
>
> Finding training classes...
> 4 classes found
> Calculating class means...
> Calculating class covariance matrices...
> Signature 1 not invertible
> Signature 2 not invertible
> Signature 3 not invertible
> Signature 4 not invertible
> i.gensig complete.
>
> My question is: do I need any package to deal with
> matrix calculus?

No, that's not needed (nor possible).

> If so, each? If not, what is wrong here?

I suspect that either the training areas are suboptimal or are
too small.

Markus