OneHotEncoder

# Список категориальных признаков
ohe_features = features_train.select_dtypes(include='object').columns.to_list()

# Список численных признаков
num_features = features_train.select_dtypes(exclude='object').columns.to_list()
num_features_to_remove = ['has_cr_card', 'is_active_member']

for item in num_features_to_remove:
    num_features.remove(item)

# Преобразуем категориальные признаки в численные
encoder_ohe = OneHotEncoder(drop='first', handle_unknown='ignore', sparse_output=False)
encoder_ohe.fit(features_train[ohe_features])

for dataset in ['features_train', 'features_valid', 'features_test']:
    vars()[dataset][encoder_ohe.get_feature_names_out()] = encoder_ohe.transform(vars()[dataset][ohe_features])
    vars()[dataset] = vars()[dataset].drop(ohe_features, axis=1)
    vars()[dataset].columns = vars()[dataset].columns.str.lower()

# Масштабирование численных признаков
scaler = StandardScaler()
scaler.fit(features_train[num_features])

for dataset in ['features_train', 'features_valid', 'features_test']:
    vars()[dataset][num_features] = scaler.transform(vars()[dataset][num_features])

for dataset in ['features_train', 'features_valid', 'features_test']:
    print('\n>>>>>>>>>> ' + dataset)
    display(vars()[dataset].head())
    print ()