sklearn Pipeline: argument of type 'ColumnTransformer' is not iterable

问题

I am attempting to use a pipeline to feed an ensemble voting classifier as I want the ensemble learner to use models that train on different feature sets. For this purpose, I followed the tutorial available at [1].

Following is the code that I could develop so far.

y = df1.index
x = preprocessing.scale(df1)

phy_features = ['A', 'B', 'C']
phy_transformer = Pipeline(steps=[('imputer', SimpleImputer(strategy='median')), ('scaler', StandardScaler())])
phy_processer = ColumnTransformer(transformers=[('phy', phy_transformer, phy_features)])

fa_features = ['D', 'E', 'F']
fa_transformer = Pipeline(steps=[('imputer', SimpleImputer(strategy='median')), ('scaler', StandardScaler())])
fa_processer = ColumnTransformer(transformers=[('fa', fa_transformer, fa_features)])


pipe_phy = Pipeline(steps=[('preprocessor', phy_processer ),('classifier', SVM)])
pipe_fa = Pipeline(steps=[('preprocessor', fa_processer ),('classifier', SVM)])

ens = VotingClassifier(estimators=[pipe_phy, pipe_fa])

cv = KFold(n_splits=10, random_state=None, shuffle=True)
for train_index, test_index in cv.split(x):
    x_train, x_test = x[train_index], x[test_index]
    y_train, y_test = y[train_index], y[test_index]
    ens.fit(x_train,y_train)
    print(ens.score(x_test, y_test))

However, when running the code, I am getting an error saying TypeError: argument of type 'ColumnTransformer' is not iterable, at the line ens.fit(x_train,y_train).

Following is the complete stack trace that I am receiving.

Traceback (most recent call last):
  File "<input>", line 1, in <module>
  File "C:\Program Files\JetBrains\PyCharm 2020.1.1\plugins\python\helpers\pydev\_pydev_bundle\pydev_umd.py", line 197, in runfile
    pydev_imports.execfile(filename, global_vars, local_vars)  # execute the script
  File "C:\Program Files\JetBrains\PyCharm 2020.1.1\plugins\python\helpers\pydev\_pydev_imps\_pydev_execfile.py", line 18, in execfile
    exec(compile(contents+"\n", file, 'exec'), glob, loc)
  File "C:/Users/ASUS/PycharmProjects/swelltest/enemble.py", line 112, in <module>
    ens.fit(x_train,y_train)
  File "C:\Users\ASUS\PycharmProjects\swelltest\venv\lib\site-packages\sklearn\ensemble\_voting.py", line 265, in fit
    return super().fit(X, transformed_y, sample_weight)
  File "C:\Users\ASUS\PycharmProjects\swelltest\venv\lib\site-packages\sklearn\ensemble\_voting.py", line 65, in fit
    names, clfs = self._validate_estimators()
  File "C:\Users\ASUS\PycharmProjects\swelltest\venv\lib\site-packages\sklearn\ensemble\_base.py", line 228, in _validate_estimators
    self._validate_names(names)
  File "C:\Users\ASUS\PycharmProjects\swelltest\venv\lib\site-packages\sklearn\utils\metaestimators.py", line 77, in _validate_names
    invalid_names = [name for name in names if '__' in name]
  File "C:\Users\ASUS\PycharmProjects\swelltest\venv\lib\site-packages\sklearn\utils\metaestimators.py", line 77, in <listcomp>
    invalid_names = [name for name in names if '__' in name]
TypeError: argument of type 'ColumnTransformer' is not iterable

Following are the values in the names list when the error is occuring.

1- ColumnTransformer(transformers=[('phy',
                                 Pipeline(steps=[('imputer',
                                                  SimpleImputer(strategy='median')),
                                                 ('scaler', StandardScaler())]),
                                 ['HR', 'RMSSD', 'SCL'])])
2- ColumnTransformer(transformers=[('fa',
                                 Pipeline(steps=[('imputer',
                                                  SimpleImputer(strategy='median')),
                                                 ('scaler', StandardScaler())]),
                                 ['Squality', 'Sneutral', 'Shappy'])])

What is the reason for this and how can I fix it?

回答1:

The estimators parameter of VotingClassifier should be a list of pairs (name, estimator), so e.g.

ens = VotingClassifier(estimators=[('phy', pipe_phy),
                                   ('fa', pipe_fa)])

(In your code, the check is trying to find the second element of the pair, hence the complaint that ColumnTransformer is not iterable.)

回答2:

I did manage to get the code to run using a workaround, which is bit ugly.

It seems like that the library is attempting to search for '__' substring of the ColumnTransform object, which it is unable to perform.

Since this name check doesn't have a significant effect on my functionality, I commented the following snippet at sklearn\utils\metaestimators.py.

invalid_names = [name for name in names if '__' in name]
if invalid_names:
    raise ValueError('Estimator names must not contain __: got '
                    '{0!r}'.format(invalid_names))

来源：https://stackoverflow.com/questions/62079006/sklearn-pipeline-argument-of-type-columntransformer-is-not-iterable

标签

python

scikit-learn

pipeline

feature-selection

ensemble-learning