search

biometrician / abe

An R package for Augmented Backward Elimination
GNU General Public License v3.0
3 stars 0 forks source link

#4: question regarding missing values #4

Closed biometrician closed 1 year ago

biometrician commented 2 years ago

In the draft it is written that "When the object data contains any missing values for any variable identified from the initial model, an error message is returned. Subjects with missing values have to be removed and the model has to be re-estimated on a revised dataset before using the functions abe or abe.resampling specifying the re-estimated model and the revised dataset. This was required due to the function abe.resampling where making resamples could in the presence of missing values lead to comparison of models which could be using datasets of different sizes."

What about automatically removing all observations with missing values in abe and abe.resampling? Additionally a warning could be given, that this was done.

This might be more convenient for the user. What do you think?

rokblagus commented 2 years ago

I don't want this to be a push the button and everything is done automatically kind of package. I think the user should (carefully) think about some things; having to remove the missing data by hand, should at least remind the user, that maybe he/she should think about this for a moment. However, if we decide to do this, it could be done (but is not as straightforward as it might seem.

biometrician commented 1 year ago

I can follow your reasoning. So I suggest to leave this at the moment.

biometrician commented 1 year ago

Close issue as not planned