This adds an informative error message to PipeOpSmote for the case in which a task with columns is passed to the pipeop that SMOTE does not know how to handle, i.e. non-feature or non-target columns..
Addtionally, we replace .select_cols(task) with task$feature_types as this should only be used when overloading .train_dt() and .predict_dt() (see documentation of PipeOpTaskPreproc).
This adds an informative error message to
PipeOpSmote
for the case in which a task with columns is passed to the pipeop that SMOTE does not know how to handle, i.e. non-feature or non-target columns..Addtionally, we replace
.select_cols(task)
withtask$feature_types
as this should only be used when overloading.train_dt()
and.predict_dt()
(see documentation ofPipeOpTaskPreproc
).closes https://github.com/mlr-org/mlr3pipelines/issues/811