All estimators should implement fit and transform, or can be ‘drop‘ or ‘passthrough‘ specifiers.

这个错误纠结了我一个小时，在网上都没找到答案，我的源代码如下，import就不贴了。

data = pd.read_csv(r'F:data\titanic\train.csv', na_values='?')

y = data['Survived']
x = data.drop(columns=['Survived','PassengerId', 'Pclass', 'Name', 'Ticket'])

x_train, x_test, y_train, y_test = train_test_split(x, y, random_state=42)

col_cat = ['Sex', 'Embarked']
col_num = ['Age', 'SibSp', 'Parch', 'Fare']

pipe_cat = make_pipeline(SimpleImputer(strategy='constant'), OneHotEncoder(handle_unknown='ignore'))
pipe_num = make_pipeline(SimpleImputer(), StandardScaler())

preprocessor = make_column_transformer((col_cat, pipe_cat), (col_num, pipe_num))

pipe = make_pipeline(preprocessor, LogisticRegression(solver='lbfgs'))

pipe.fit(x_train, y_train)
accuracy = pipe.score(x_test, y_test)
print('Accuracy score of the {} is {:.2f}'.format(pipe.__class__.__name__, accuracy))

编译的时候他非要报错，下面的错误：

Traceback (most recent call last):
  File "E:\Anaconda\install\envs\pytorch\lib\site-packages\sklearn\pipeline.py", line 390, in fit
    Xt = self._fit(X, y, **fit_params_steps)
  File "E:\Anaconda\install\envs\pytorch\lib\site-packages\sklearn\pipeline.py", line 355, in _fit
    **fit_params_steps[name],
  File "E:\Anaconda\install\envs\pytorch\lib\site-packages\joblib\memory.py", line 349, in __call__
    return self.func(*args, **kwargs)
  File "E:\Anaconda\install\envs\pytorch\lib\site-packages\sklearn\pipeline.py", line 893, in _fit_transform_one
    res = transformer.fit_transform(X, y, **fit_params)
  File "E:\Anaconda\install\envs\pytorch\lib\site-packages\sklearn\compose\_column_transformer.py", line 671, in fit_transform
    self._validate_transformers()
  File "E:\Anaconda\install\envs\pytorch\lib\site-packages\sklearn\compose\_column_transformer.py", line 339, in _validate_transformers
    "specifiers. '%s' (type %s) doesn't." % (t, type(t))
TypeError: All estimators should implement fit and transform, or can be 'drop' or 'passthrough' specifiers. '['Sex', 'Embarked']' (type <class 'list'>) doesn't.

结果傻逼的看了一下官方的 API 文档才发现， make_column_transformer 这个管道连接其必须的 tuple 中需要管道在前，列名在后，所以只需要将这俩位置换一下就行

preprocessor = make_column_transformer((pipe_cat, col_cat), (pipe_num, col_num))

气死老娘了！