来自 sklearn-template 文档：

请注意，转换仅将 X 作为输入，并应返回 X 的转换版本

所以现在在你的类的方法中，数据帧被替换为transformLabelYEncoderXy_encoded'numpy.ndarray'

看起来您正在尝试为预处理管道构建一些自定义类。我不知道您使用的是什么版本，但之前的任何内容都会尝试将输入转换为.set_output方法更改了此行为。scikit-learn1.2.Xnp.ndarray

scikit-learn1.3.2

如果您能够升级，则您引发的错误将更改为：

class Cleaner(BaseEstimator, TransformerMixin):
    def fit(self, X, y=None):
        return self

    def transform(self, X):
        X = X.drop(["Name", "PassengerId", "FoodCourt", "ShoppingMall", "Cabin"], axis=1)
        return X

class ObjectEncoder(BaseEstimator, TransformerMixin):
    def fit(self, X, y=None):
        return self

    def transform(self, X):
        columns_to_encode = ["HomePlanet", "CryoSleep", "Destination", "VIP"]
        encoder = OneHotEncoder()
        # Encode each attribute and add the columns to the DataFrame
        for column in columns_to_encode:
            matrix = encoder.fit_transform(X[[column]]).toarray()
            column_names = [f"{column}_{i}" for i in range(matrix.shape[1])]
            for i in range(len(matrix.T)):
                X[column_names[i]] = matrix.T[i]

        # Remove old parameters
        X = X.drop(["HomePlanet", "CryoSleep", "Destination", "VIP"], axis=1)
        return X

class LabelYEncoder(BaseEstimator, TransformerMixin):
    def fit(self, X, y=None):
        return self

    def transform(self, X):
        label_encoder = LabelEncoder()
        y_encoded = label_encoder.fit_transform(y)
        return y_encoded

class NullImputer(BaseEstimator, TransformerMixin):
    def fit(self, X, y=None):
        return self

    def transform(self, X):
        features_with_missing_values = X.columns[X.isnull().any()].tolist()
        imputer = SimpleImputer(strategy="median")
        X[features_with_missing_values] = imputer.fit_transform(X[features_with_missing_values])
        return X

pipeline = Pipeline(
    steps=[
        ("cleaner", Cleaner()),
        ("object_encoder", ObjectEncoder()),
        ("label_y_encoder", LabelYEncoder()),
        ("null_imputer", NullImputer()),
    ]
)

df = pd.read_csv("train.csv")

# "In the code below X is the same dataset after removing the Transformed Feature from it."
# Idk what "Transformed Feature" is, but I assume you mean "Transported"
strat_train_set = df.drop(columns="Transported")

X = pipeline.fit_transform(strat_train_set)

---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
Cell In[21], line 61
     57 # "In the code below X is the same dataset after removing the Transformed Feature from it."
     58 # Idk what "Transformed Feature" is, but I assume you mean "Transported"
     59 strat_train_set = df.drop(columns="Transported")
---> 61 X = pipeline.fit_transform(strat_train_set)

File ~\Sandbox\PRIVATE-RAI\env\Lib\site-packages\sklearn\base.py:1152, in _fit_context.<locals>.decorator.<locals>.wrapper(estimator, *args, **kwargs)
   1145     estimator._validate_params()
   1147 with config_context(
   1148     skip_parameter_validation=(
   1149         prefer_skip_nested_validation or global_skip_validation
   1150     )
   1151 ):
-> 1152     return fit_method(estimator, *args, **kwargs)

File ~\Sandbox\PRIVATE-RAI\env\Lib\site-packages\sklearn\pipeline.py:471, in Pipeline.fit_transform(self, X, y, **fit_params)
    444 """Fit the model and transform with the final estimator.
    445 
    446 Fits all the transformers one after the other and transform the
   (...)
    468     Transformed samples.
    469 """
    470 fit_params_steps = self._check_fit_params(**fit_params)
--> 471 Xt = self._fit(X, y, **fit_params_steps)
    473 last_step = self._final_estimator
    474 with _print_elapsed_time("Pipeline", self._log_message(len(self.steps) - 1)):

File ~\Sandbox\PRIVATE-RAI\env\Lib\site-packages\sklearn\pipeline.py:377, in Pipeline._fit(self, X, y, **fit_params_steps)
    375     cloned_transformer = clone(transformer)
    376 # Fit or load from cache the current transformer
--> 377 X, fitted_transformer = fit_transform_one_cached(
    378     cloned_transformer,
    379     X,
    380     y,
    381     None,
    382     message_clsname="Pipeline",
    383     message=self._log_message(step_idx),
    384     **fit_params_steps[name],
    385 )
    386 # Replace the transformer of the step with the fitted
    387 # transformer. This is necessary when loading the transformer
    388 # from the cache.
    389 self.steps[step_idx] = (name, fitted_transformer)

File ~\Sandbox\PRIVATE-RAI\env\Lib\site-packages\joblib\memory.py:353, in NotMemorizedFunc.__call__(self, *args, **kwargs)
    352 def __call__(self, *args, **kwargs):
--> 353     return self.func(*args, **kwargs)

File ~\Sandbox\PRIVATE-RAI\env\Lib\site-packages\sklearn\pipeline.py:957, in _fit_transform_one(transformer, X, y, weight, message_clsname, message, **fit_params)
    955 with _print_elapsed_time(message_clsname, message):
    956     if hasattr(transformer, "fit_transform"):
--> 957         res = transformer.fit_transform(X, y, **fit_params)
    958     else:
    959         res = transformer.fit(X, y, **fit_params).transform(X)

File ~\Sandbox\PRIVATE-RAI\env\Lib\site-packages\sklearn\utils\_set_output.py:157, in _wrap_method_output.<locals>.wrapped(self, X, *args, **kwargs)
    155 @wraps(f)
    156 def wrapped(self, X, *args, **kwargs):
--> 157     data_to_wrap = f(self, X, *args, **kwargs)
    158     if isinstance(data_to_wrap, tuple):
    159         # only wrap the first output for cross decomposition
    160         return_tuple = (
    161             _wrap_data_with_container(method, data_to_wrap[0], X, self),
    162             *data_to_wrap[1:],
    163         )

File ~\Sandbox\PRIVATE-RAI\env\Lib\site-packages\sklearn\base.py:916, in TransformerMixin.fit_transform(self, X, y, **fit_params)
    912 # non-optimized default implementation; override when a better
    913 # method is possible for a given clustering algorithm
    914 if y is None:
    915     # fit method of arity 1 (unsupervised transformation)
--> 916     return self.fit(X, **fit_params).transform(X)
    917 else:
    918     # fit method of arity 2 (supervised transformation)
    919     return self.fit(X, y, **fit_params).transform(X)

File ~\Sandbox\PRIVATE-RAI\env\Lib\site-packages\sklearn\utils\_set_output.py:157, in _wrap_method_output.<locals>.wrapped(self, X, *args, **kwargs)
    155 @wraps(f)
    156 def wrapped(self, X, *args, **kwargs):
--> 157     data_to_wrap = f(self, X, *args, **kwargs)
    158     if isinstance(data_to_wrap, tuple):
    159         # only wrap the first output for cross decomposition
    160         return_tuple = (
    161             _wrap_data_with_container(method, data_to_wrap[0], X, self),
    162             *data_to_wrap[1:],
    163         )

Cell In[21], line 33, in LabelYEncoder.transform(self, X)
     31 def transform(self, X):
     32     label_encoder = LabelEncoder()
---> 33     y_encoded = label_encoder.fit_transform(y)
     34     return y_encoded

NameError: name 'y' is not defined

您可以通过编辑 ：method： 来接受参数来解决这个问题。但就我个人而言，我会修改您的自定义类以使其更通用，并使用 a 来控制每个步骤的输入。如下所示：LabelYEncoder.transformyColumnTransformer

import pandas as pd
from sklearn.compose import ColumnTransformer
from sklearn.impute import SimpleImputer
from sklearn.pipeline import FeatureUnion, Pipeline
from sklearn.preprocessing import FunctionTransformer, OneHotEncoder, LabelEncoder

# Separating `X` and `y`
df = pd.read_csv("train.csv")
X = df.drop(columns="Transported")
y = df.Transported

cols_to_drop = ["Name", "PassengerId", "FoodCourt", "ShoppingMall", "Cabin"]
cols_to_one_hot = ["HomePlanet", "CryoSleep", "Destination", "VIP"]
# Remove the `cols_to_drop` and `cols_to_one_hot`.
# The remainder will be passed through via the identity function.
cols_to_keep = X.drop(columns=[*cols_to_drop, *cols_to_one_hot]).columns

prep_trf = ColumnTransformer(
    transformers=[
        # One hot encode the `cols_to_one_hot`
        ("one_hot", OneHotEncoder(sparse_output=False), cols_to_one_hot),
        # Impute the `cols_to_keep`
        ("imputer", SimpleImputer(strategy="median"), cols_to_keep),
    ],
    # Drop the remaining columns not mentioned in 
    # `cols_to_one_hot` or `cols_to_keep`
    remainder="drop",
    verbose_feature_names_out=False,
    # Transform output to `pandas.DataFrame` (optional)
).set_output(transform="pandas")

# Apply the transformations to `X`.
X = prep_trf.fit_transform(X)

调用将处理输入要素的所有预处理。因为只接受输入，不支持 ：method：，所以你不能把它添加到对象中。事实上，从技术上讲，如果没有一些自定义代码，就无法将其添加到管道中。相反，您应该独立转换变量，或者完全忽略它（因为它已经是具有布尔值的二进制形式）。prep_trf.fit_transformXLabelEncoderyset_outputprep_trfy

scikit-learn1.1.3 （< 1.2.X）

import pandas as pd
from sklearn.compose import ColumnTransformer
from sklearn.impute import SimpleImputer
from sklearn.pipeline import FeatureUnion, Pipeline
from sklearn.preprocessing import FunctionTransformer, OneHotEncoder, LabelEncoder

# Separating `X` and `y`
df = pd.read_csv("train.csv")
X = df.drop(columns="Transported")
y = df.Transported

cols_to_drop = ["Name", "PassengerId", "FoodCourt", "ShoppingMall", "Cabin"]
cols_to_one_hot = ["HomePlanet", "CryoSleep", "Destination", "VIP"]
# Remove the `cols_to_drop` and `cols_to_one_hot`.
# The remainder will be passed through via the identity function.
cols_to_keep = X.drop(columns=[*cols_to_drop, *cols_to_one_hot]).columns

prep_trf = ColumnTransformer(
    transformers=[
        # One hot encode the `cols_to_one_hot`
        # NOTE: had to change the `sparse_output` parameter to `sparse`.
        ("one_hot", OneHotEncoder(sparse=False), cols_to_one_hot),
        # Impute the `cols_to_keep`
        ("imputer", SimpleImputer(strategy="median"), cols_to_keep),
    ],
    # Drop the remaining columns not mentioned in 
    # `cols_to_one_hot` or `cols_to_keep`
    remainder="drop",
    verbose_feature_names_out=False,
    # NOTE: :method:`set_output` does not exist in scikit-learn version < 1.2.X
)

# Apply the transformations to `X`.
X_ = prep_trf.fit_transform(X)

因为 1.1.3 版本没有 ：method：，我们需要重新创建自己：scikit-learnset_outputpandas.DataFrame

X_v1_1_3 = pd.DataFrame(
    data=X_,
    # Track the index just incase it was modified earlier.
    # This can affect `joins`.
    index=X.index,
    # Use the :method:`prep_trf_get_feature_names_out` to get the 
    # correct column names.
    columns=prep_trf.get_feature_names_out(),
)

在这两种情况下，输出将是相同的。在两个版本中分别应用 to。LabelEncodery

其他资源：

• 1.1.TransformerMixin 的 X 实现
• 1.2.TransformerMixin 的实现
• set_output API 简介