I think a basic functionality that the package should have built in is basic imputation. Imputing based on mean, median or some other simply calculated single value. This would primarily be useful for measurement values. Would also be nice to have a threshold for including a certain measurement, i.e. if missing values are more than 50% don't include them. Or a threshold based on absolute number of measurements, since otherwise the imputation would be to noisy.
We have today the new age stratified imputation but I think that is maybe more for advanced use cases.
This should be very straightforward to add using the featureEngineering api.
I think a basic functionality that the package should have built in is basic imputation. Imputing based on
mean,medianor some other simply calculated single value. This would primarily be useful for measurement values. Would also be nice to have a threshold for including a certain measurement, i.e. if missing values are more than 50% don't include them. Or a threshold based on absolute number of measurements, since otherwise the imputation would be to noisy.We have today the new age stratified imputation but I think that is maybe more for advanced use cases.
This should be very straightforward to add using the
featureEngineeringapi.