Nu skal du sammenligne de modeller, vi har gennemgået, ved hjælp af ét datasæt — **breast cancer dataset**. Målvariablen er kolonnen `'diagnosis'`, hvor `1` repræsenterer maligne og `0` repræsenterer benigne tilfælde.

Du skal anvende `GridSearchCV` på hver model for at finde de bedste parametre. I denne opgave skal du bruge **recall** som evalueringsmetrik, da **minimering af falske negativer** er afgørende. For at få `GridSearchCV` til at vælge de bedste parametre baseret på recall, skal du sætte `scoring='recall'`.

import unittest
import importlib


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 TestUserCode(unittest.TestCase):

    def test_knn_params_is_correct(self):
        import user_code

        expected_value = {'n_neighbors': [3, 5, 7, 12]}

        variable = 'knn_params'
        actual_value = getattr(user_code, variable, None)
        if actual_value is None:
            condition = False
            failure_message = f"The `{variable}` variable is not declared."
        elif isinstance(actual_value, dict):
            condition = actual_value == expected_value
            failure_message = f"Expected `{variable}` to contain `{expected_value}`, but got `{actual_value}`."
        else:
            condition = False
            failure_message = f"`{variable}` is not a `dict`."

        _dynamic_test(
            self,
            condition,
            f"`{variable}` contains the correct values.",
            failure_message
        )

    def test_lr_params_is_correct(self):
        import user_code

        expected_value = {'C': [0.1, 1, 10]}

        variable = 'lr_params'
        actual_value = getattr(user_code, variable, None)
        if actual_value is None:
            condition = False
            failure_message = f"The `{variable}` variable is not declared."
        elif isinstance(actual_value, dict):
            condition = actual_value == expected_value
            failure_message = f"Expected `{variable}` to contain `{expected_value}`, but got `{actual_value}`."
        else:
            condition = False
            failure_message = f"`{variable}` is not a `dict`."

        _dynamic_test(
            self,
            condition,
            f"`{variable}` contains the correct values.",
            failure_message
        )

    def test_dt_params_is_correct(self):
        import user_code

        expected_value = {'max_depth': [2, 4, 6, 10], 'min_samples_leaf': [1, 2, 4, 7]}

        variable = 'dt_params'
        actual_value = getattr(user_code, variable, None)
        if actual_value is None:
            condition = False
            failure_message = f"The `{variable}` variable is not declared."
        elif isinstance(actual_value, dict):
            condition = actual_value == expected_value
            failure_message = f"Expected `{variable}` to contain `{expected_value}`, but got `{actual_value}`."
        else:
            condition = False
            failure_message = f"`{variable}` is not a `dict`."

        _dynamic_test(
            self,
            condition,
            f"`{variable}` contains the correct values.",
            failure_message
        )

    def test_rf_params_is_correct(self):
        import user_code

        expected_value = {'max_depth': [2, 4, 6], 'n_estimators': [20, 50, 100]}

        variable = 'rf_params'
        actual_value = getattr(user_code, variable, None)
        if actual_value is None:
            condition = False
            failure_message = f"The `{variable}` variable is not declared."
        elif isinstance(actual_value, dict):
            condition = actual_value == expected_value
            failure_message = f"Expected `{variable}` to contain `{expected_value}`, but got `{actual_value}`."
        else:
            condition = False
            failure_message = f"`{variable}` is not a `dict`."

        _dynamic_test(
            self,
            condition,
            f"`{variable}` contains the correct values.",
            failure_message
        )

    def test_knn_grid_is_correct(self):
        import user_code
        from sklearn.model_selection import GridSearchCV
        from sklearn.neighbors import KNeighborsClassifier

        param_grid = {'n_neighbors': [3, 5, 7, 12]}

        variable = 'knn_grid'
        actual_value = getattr(user_code, variable, None)
        if actual_value is None:
            condition = False
            failure_message = f"The `{variable}` variable is not declared."
        elif isinstance(actual_value, GridSearchCV):
            condition = isinstance(actual_value.estimator,
                                   KNeighborsClassifier) and actual_value.param_grid == param_grid
            failure_message = f"Expected `{variable}` to be a `GridSearchCV` with `estimator=KNeighborsClassifier()`, `param_grid={param_grid}`, but got `estimator={actual_value.estimator}`, `param_grid={actual_value.param_grid}`."
        else:
            condition = False
            failure_message = f"`{variable}` is not a `GridSearchCV`."

        _dynamic_test(
            self,
            condition,
            f"`{variable}` is a `GridSearchCV` with `estimator=knn` and `param_grid=knn_grid`.",
            failure_message
        )

    def test_lr_grid_is_correct(self):
        import user_code
        from sklearn.model_selection import GridSearchCV
        from sklearn.linear_model import LogisticRegression

        param_grid = {'C': [0.1, 1, 10]}

        variable = 'lr_grid'
        actual_value = getattr(user_code, variable, None)
        if actual_value is None:
            condition = False
            failure_message = f"The `{variable}` variable is not declared."
        elif isinstance(actual_value, GridSearchCV):
            condition = isinstance(actual_value.estimator,
                                   LogisticRegression) and actual_value.param_grid == param_grid
            failure_message = f"Expected `{variable}` to be a `GridSearchCV` with `estimator=LogisticRegression()`, `param_grid={param_grid}`, but got `estimator={actual_value.estimator}`, `param_grid={actual_value.param_grid}`."
        else:
            condition = False
            failure_message = f"`{variable}` is not a `GridSearchCV`."

        _dynamic_test(
            self,
            condition,
            f"`{variable}` is a `GridSearchCV` with `estimator=lr` and `param_grid=lr_grid`.",
            failure_message
        )

    def test_dt_grid_is_correct(self):
        import user_code
        from sklearn.model_selection import GridSearchCV
        from sklearn.tree import DecisionTreeClassifier

        param_grid = {'max_depth': [2, 4, 6, 10], 'min_samples_leaf': [1, 2, 4, 7]}

        variable = 'dt_grid'
        actual_value = getattr(user_code, variable, None)
        if actual_value is None:
            condition = False
            failure_message = f"The `{variable}` variable is not declared."
        elif isinstance(actual_value, GridSearchCV):
            condition = isinstance(actual_value.estimator,
                                   DecisionTreeClassifier) and actual_value.param_grid == param_grid
            failure_message = f"Expected `{variable}` to be a `GridSearchCV` with `estimator=DecisionTreeClassifier()`, `param_grid={param_grid}`, but got `estimator={actual_value.estimator}`, `param_grid={actual_value.param_grid}`."
        else:
            condition = False
            failure_message = f"`{variable}` is not a `GridSearchCV`."

        _dynamic_test(
            self,
            condition,
            f"`{variable}` is a `GridSearchCV` with `estimator=dt` and `param_grid=dt_grid`.",
            failure_message
        )

    def test_rf_grid_is_correct(self):
        import user_code
        from sklearn.model_selection import GridSearchCV
        from sklearn.ensemble import RandomForestClassifier

        param_grid = {'max_depth': [2, 4, 6], 'n_estimators': [20, 50, 100]}

        variable = 'rf_grid'
        actual_value = getattr(user_code, variable, None)
        if actual_value is None:
            condition = False
            failure_message = f"The `{variable}` variable is not declared."
        elif isinstance(actual_value, GridSearchCV):
            condition = isinstance(actual_value.estimator,
                                   RandomForestClassifier) and actual_value.param_grid == param_grid
            failure_message = f"Expected `{variable}` to be a `GridSearchCV` with `estimator=RandomForestClassifier()`, `param_grid={param_grid}`, but got `estimator={actual_value.estimator}`, `param_grid={actual_value.param_grid}`."
        else:
            condition = False
            failure_message = f"`{variable}` is not a `GridSearchCV`."

        _dynamic_test(
            self,
            condition,
            f"`{variable}` is a `GridSearchCV` with `estimator=rf` and `param_grid=rf_grid`.",
            failure_message
        )


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

test_main.py

Behersk de centrale klassifikationsalgoritmer, der driver moderne maskinlæring. Udforsk hvordan modeller som k-NN, logistisk regression, beslutningstræer og random forests foretager forudsigelser, evaluerer deres nøjagtighed, og forstå hvornår hver model bør anvendes. Opbyg færdigheder til at sammenligne modeller og vælge den bedste til dine data.

Opdag, hvordan k-nærmeste naboer-algoritmen foretager forudsigelser baseret på lighed. Lær at håndtere flere egenskaber, justere parametre og anvende krydsvalidering for at forbedre nøjagtigheden.

Forstå, hvordan logistisk regression modellerer sandsynligheder og klassificerer udfald. Øv implementering, fortolkning af beslutningsgrænser og anvendelse af regularisering for at forhindre overfitting.

Lær, hvordan beslutningstræer opdeler data i meningsfulde grupper baseret på featureværdier. Udforsk, hvordan parametre som trædybde og minimum antal prøver per blad påvirker modellens ydeevne og generalisering.

Undersøg, hvordan random forests kombinerer flere beslutningstræer for at forbedre nøjagtighed og robusthed. Forstå betydningen af tilfældighed og anvend denne ensemble-metode på virkelige data.

Evaluering af modeller ved hjælp af metrikker såsom nøjagtighed, præcision, recall og F1-score. Lær at fortolke forvekslingsmatrixer og sammenligne flere klassifikatorer for at identificere den bedst præsterende model.

Udfordring: Sammenligning af Modeller

Løsning

Awesome!

Udfordring: Sammenligning af Modeller

Løsning