Om de vorige drie hoofdstukken samen te vatten, volgt hier een tabel die aangeeft welke **encoder** gebruikt moet worden:

In deze uitdaging werk je met de **penguins dataset** (geen ontbrekende waarden). Alle categorische kenmerken — inclusief de target `'species'` — moeten worden gecodeerd voor gebruik in ML.

import pandas as pd

df = pd.read_csv('https://codefinity-content-media.s3.eu-west-1.amazonaws.com/a65bbc96-309e-4df9-a790-a1eb8c815a1c/penguins_imputed.csv')

print(df.head())

Houd er rekening mee dat `'island'` en `'sex'` categorische **kenmerken** zijn en `'species'` een categorisch **doel** is.

import unittest
import pandas as pd
import numpy as np

def _dynamic_test(test_case, condition, success_message, failure_message):
    if condition:
        test_case._testMethodName = success_message
        test_case.assertTrue(True, success_message)
    else:
        test_case._testMethodName = failure_message
        test_case.fail(failure_message)

class TestEncoding(unittest.TestCase):

    @classmethod
    def setUpClass(cls):
        cls.df_raw = pd.read_csv(
            'https://codefinity-content-media.s3.eu-west-1.amazonaws.com/a65bbc96-309e-4df9-a790-a1eb8c815a1c/penguins_imputed.csv'
        )
        import user_code
        cls.user_code = user_code

    def test_imports_encoders(self):
        from sklearn.preprocessing import OneHotEncoder, LabelEncoder
        uc = self.user_code
        cond = isinstance(getattr(uc, 'feature_enc', None), OneHotEncoder) and \
               isinstance(getattr(uc, 'label_enc', None), LabelEncoder)
        _dynamic_test(
            self,
            cond,
            "Used OneHotEncoder for features and LabelEncoder for target",
            "Used OneHotEncoder for features and LabelEncoder for target"
        )

    def test_feature_columns_encoded(self):
        uc = self.user_code
        expected = ['island_Biscoe', 'island_Dream', 'island_Torgersen',
                    'sex_FEMALE', 'sex_MALE']
        cond = all(col in uc.X.columns for col in expected)
        _dynamic_test(
            self,
            cond,
            "Encoded columns were added to X",
            "Encoded columns were added to X"
        )

    def test_original_columns_dropped(self):
        uc = self.user_code
        cond = ('island' not in uc.X.columns) and ('sex' not in uc.X.columns)
        _dynamic_test(
            self,
            cond,
            "Original categorical columns were dropped",
            "Original categorical columns were dropped"
        )

    def test_y_encoded(self):
        uc = self.user_code
        y_arr = np.asarray(uc.y)
        is_integer_dtype = np.issubdtype(y_arr.dtype, np.integer)
        le_fitted = hasattr(getattr(uc, 'label_enc', object()), 'classes_')
        cond = (y_arr.ndim == 1) and is_integer_dtype and le_fitted
        _dynamic_test(
            self,
            cond,
            "Target y must be encoded into integers using LabelEncoder",
            "Target y must be encoded into integers using LabelEncoder"
        )

if __name__ == "__main__":
    unittest.main()


test_code.py

Beheers de basisprincipes van Machine Learning en de Scikit-learn-bibliotheek. Verken de volledige ML-werkstroom, van het omgaan met ontbrekende waarden en het coderen van categorische gegevens tot het schalen van kenmerken. Bouw efficiënte, lekvrije gegevensvoorverwerkingspijplijnen met behulp van ColumnTransformer. Zet ruwe datasets om in modelklare structuren en implementeer robuuste voorspellende pijplijnen.

Uitdaging: Coderen van Categorische Variabelen

Oplossing