Say we’ve got a dataframe with a mixture of categorical and numerical features which will be used for binary classification with missing values.