Currently, we are only creating a Classifier performance report output, which is calculated from the training sample via cross-validation (or soon via OOB, see #910).
It is proposed to also calculate a Classification layer accuracy and area report output, using the Classification layer output as stratification.
Unfortunately, we can't do this straight away, because the Training dataset input isn't aware of the sample locations anymore (wait for #911).
Discussed with @d-pflugmacher.
Currently, we are only creating a Classifier performance report output, which is calculated from the training sample via cross-validation (or soon via OOB, see #910).
It is proposed to also calculate a Classification layer accuracy and area report output, using the Classification layer output as stratification.
Unfortunately, we can't do this straight away, because the Training dataset input isn't aware of the sample locations anymore (wait for #911).