Thread Subject:

Combine dataset groups in grpstats()

I have a dataset like the following which I'd like to group the stats into {'test+training', 'test'} instead of {'training', 'test'} with grpstats().

 DD = dataset( {nominal({'test'; 'training'; 'test'; 'test'; 'training'}), 'classification'}, {[0.3;0.5;0.7;0.9;1.1], 'value'} )

Is there a neat way to do it for overlapping groups?

> I have a dataset like the following which I'd like to group the stats into
> {'test+training', 'test'} instead of {'training', 'test'} with grpstats().
> DD = dataset( {nominal({'test'; 'training'; 'test'; 'test'; 'training'}),
> 'classification'}, {[0.3;0.5;0.7;0.9;1.1], 'value'} )
>
> Is there a neat way to do it for overlapping groups?

If the computing time is not excessive, you may find it makes best use of
your own time to compute the statistics once for the full data set, then
again for the two groups.

-- Tom