vanderschaarlab / autoprognosis

A system for automating the design of predictive modeling pipelines tailored for clinical prognosis.
https://www.autoprognosis.vanderschaar-lab.com/
Apache License 2.0
119 stars 27 forks source link

"Let the optimizer find the optimal imputer" in "Classification AutoML with imputation" cannot recognize NaN at first #74

Open williamty opened 1 year ago

williamty commented 1 year ago

If don't predefine imputer, "let the optimizer find the optimal imputer" in "Classification AutoML with imputation", the optimal imputer will encounter a problem that it "does not accept missing values encoded as NaN natively". Shouldn't it first judge whether there is null data in the dataset? Otherwise what's the meaning of "optimal"?