i think this can be solved by filling the missing values first right after loading the data

 data['variable'].fillna('No data')

and turn them into categorical

data['variable_q'] = pd.Categorical(data['variable'])