Is there a ‘good’ imputer for the categorical variable. The default/available SimpleImputer within sklearn only allows (strategy=‘constant’) unline mean. median etc. While other classifiers will allow NaN but is there any available method that can be used. ( rather than using pandas to write up a method for imputing them)
Most common is another SimpleImputer strategy, which works for categorical features. Though, I usually fill in missing categorical features with ‘missing’ as that allows your model to treat missing values differently which could help improve performance.