[Model Designer] Predict classification layer: make classifier argument accept model outputs as input

[Fabian Thiel]

As the titel suggests I currently can not find an option to make the Classifier argument be an output of a previous algorithm (e.g. Fit RandomForestClassifier).

If my model looks like this, I cannot bridge the Fit and the Predict stage:

In the Fit algorithm for example I have several options to populate the parameter(s) (in this example Training dataset). The same would be nice for the Classifier argument in the Predict classification algorithm:

‌

[Andreas Rabe]

That is related to a QGIS bug: https://github.com/qgis/QGIS/issues/48816

Unfortunately, the API classes I used are already deprecated, so the bug won’t be fixed.

I already talked with @Benjamin Jakimow about it, and Benjamin has an idea how to change the implementation.

[Andreas Rabe]

Duplicate of #1366.