What's the difference between morfist and the sklearn RandomForestRegressor with multi-output when dealing with only regression problems??

[Qianqian-Yang]

When dealing with only regression problems, the results of sklearn RandomForestRegressor is different from that of morfist, what caused the difference? I'm not quite clear about the implementary details of multioutput sklearn RandomForestRegressor and morfist when conducting multioutput-regression.

[donlnz]

Specifically for regression problems, there's no substantial difference to scikit-learn's RandomForestRegressor that I'm aware of. Same goes for classification problems and scikit-learn's RandomForestClassifier.

What morfist does differently (and, I don't think there's currently support for this in scikit-learn) is offer random forest models with a mix of classification and regression tasks. By default, morfist treats output variables as regression variables, but you can pass a list of output-variable indices to the class_targets parameter of the constructor to force the training algorithm to evaluate select output variables as discrete/classification targets.