Encoding categorical features

encoding.py

import category_encoders as ce

#Ordinal Encoding
ord_enc = ce.OrdinalEncoder()
df[['XXX', 'YYY', 'ZZZ']] = ord_enc.fit_transform(
    df[['XXX', 'YYY', 'ZZZ']])

# OneHotEncoding
ohe_columns = ['XXX', 'YYY', 'ZZZ']
df = pd.get_dummies(df, columns = ohe_columns)

# Binary Encoding
bin_encoder = ce.BinaryEncoder(cols = ['XXX', 'YYY', 'ZZZ'])
df_bin = bin_encoder.fit_transform(df[['XXX', 'YYY', 'ZZZ']])
df = pd.concat([df, df_bin], axis = 1)

Use OneHotEncoding if number of unique characteristics is 15 - 20, not more than 20.
Use BinaryEncoding if number of unique characteristics is exceeding 20.