oneachcolumn and onsubframe want dataframe output from their transformers. if the method exists, they call set_output on the transformer so that scikit-learn transformers produce dataframe output. however set_output may fail in cases where the output would be correct anyway, eg a Pipeline containing transformers that produce dataframes by default but not exposing a set_output method.
this pr allows set_output to fail and still attempts the transformation -- the type of the output is checked in any case
oneachcolumn and onsubframe want dataframe output from their transformers. if the method exists, they call
set_outputon the transformer so that scikit-learn transformers produce dataframe output. howeverset_outputmay fail in cases where the output would be correct anyway, eg a Pipeline containing transformers that produce dataframes by default but not exposing aset_outputmethod.this pr allows
set_outputto fail and still attempts the transformation -- the type of the output is checked in any case