Cross_val_score无法与roc_auc和multiclass一起使用。

Question

我想做什么：

我希望对一个多类问题使用roc_auc计算cross_val_score。

我尝试了什么：

这是一个使用鸢尾花数据集创建的可复现示例。

from sklearn.datasets import load_iris
from sklearn.preprocessing import OneHotEncoder
from sklearn.model_selection import cross_val_score  
iris = load_iris()
X = pd.DataFrame(data=iris.data, columns=iris.feature_names)

我对目标进行独热编码

encoder = OneHotEncoder()
y = encoder.fit_transform(pd.DataFrame(iris.target)).toarray()

我使用一种决策树分类器

model = DecisionTreeClassifier(max_depth=1)

最终我执行了交叉验证

cross_val_score(model, X, y, cv=3, scoring="roc_auc")

问题出在哪里：

这最后一行抛出以下错误

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-87-91dc6fa67512> in <module>()
----> 1 cross_val_score(model, X, y, cv=3, scoring="roc_auc")

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_validation.py in cross_val_score(estimator, X, y, groups, scoring, cv, n_jobs, verbose, fit_params, pre_dispatch)
    340                                 n_jobs=n_jobs, verbose=verbose,
    341                                 fit_params=fit_params,
--> 342                                 pre_dispatch=pre_dispatch)
    343     return cv_results['test_score']
    344 

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_validation.py in cross_validate(estimator, X, y, groups, scoring, cv, n_jobs, verbose, fit_params, pre_dispatch, return_train_score)
    204             fit_params, return_train_score=return_train_score,
    205             return_times=True)
--> 206         for train, test in cv.split(X, y, groups))
    207 
    208     if return_train_score:

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/externals/joblib/parallel.py in __call__(self, iterable)
    777             # was dispatched. In particular this covers the edge
    778             # case of Parallel used with an exhausted iterator.
--> 779             while self.dispatch_one_batch(iterator):
    780                 self._iterating = True
    781             else:

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/externals/joblib/parallel.py in dispatch_one_batch(self, iterator)
    623                 return False
    624             else:
--> 625                 self._dispatch(tasks)
    626                 return True
    627 

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/externals/joblib/parallel.py in _dispatch(self, batch)
    586         dispatch_timestamp = time.time()
    587         cb = BatchCompletionCallBack(dispatch_timestamp, len(batch), self)
--> 588         job = self._backend.apply_async(batch, callback=cb)
    589         self._jobs.append(job)
    590 

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/externals/joblib/_parallel_backends.py in apply_async(self, func, callback)
    109     def apply_async(self, func, callback=None):
    110         """Schedule a func to be run"""
--> 111         result = ImmediateResult(func)
    112         if callback:
    113             callback(result)

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/externals/joblib/_parallel_backends.py in __init__(self, batch)
    330         # Don't delay the application, to avoid keeping the input
    331         # arguments in memory
--> 332         self.results = batch()
    333 
    334     def get(self):

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/externals/joblib/parallel.py in __call__(self)
    129 
    130     def __call__(self):
--> 131         return [func(*args, **kwargs) for func, args, kwargs in self.items]
    132 
    133     def __len__(self):

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/externals/joblib/parallel.py in <listcomp>(.0)
    129 
    130     def __call__(self):
--> 131         return [func(*args, **kwargs) for func, args, kwargs in self.items]
    132 
    133     def __len__(self):

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_validation.py in _fit_and_score(estimator, X, y, scorer, train, test, verbose, parameters, fit_params, return_train_score, return_parameters, return_n_test_samples, return_times, error_score)
    486         fit_time = time.time() - start_time
    487         # _score will return dict if is_multimetric is True
--> 488         test_scores = _score(estimator, X_test, y_test, scorer, is_multimetric)
    489         score_time = time.time() - start_time - fit_time
    490         if return_train_score:

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_validation.py in _score(estimator, X_test, y_test, scorer, is_multimetric)
    521     """
    522     if is_multimetric:
--> 523         return _multimetric_score(estimator, X_test, y_test, scorer)
    524     else:
    525         if y_test is None:

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/model_selection/_validation.py in _multimetric_score(estimator, X_test, y_test, scorers)
    551             score = scorer(estimator, X_test)
    552         else:
--> 553             score = scorer(estimator, X_test, y_test)
    554 
    555         if hasattr(score, 'item'):

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/metrics/scorer.py in __call__(self, clf, X, y, sample_weight)
    204                                                  **self._kwargs)
    205         else:
--> 206             return self._sign * self._score_func(y, y_pred, **self._kwargs)
    207 
    208     def _factory_args(self):

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/metrics/ranking.py in roc_auc_score(y_true, y_score, average, sample_weight)
    275     return _average_binary_score(
    276         _binary_roc_auc_score, y_true, y_score, average,
--> 277         sample_weight=sample_weight)
    278 
    279 

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/metrics/base.py in _average_binary_score(binary_metric, y_true, y_score, average, sample_weight)
    116         y_score_c = y_score.take([c], axis=not_average_axis).ravel()
    117         score[c] = binary_metric(y_true_c, y_score_c,
--> 118                                  sample_weight=score_weight)
    119 
    120     # Average the results

~/programs/anaconda3/lib/python3.7/site-packages/sklearn/metrics/ranking.py in _binary_roc_auc_score(y_true, y_score, sample_weight)
    266     def _binary_roc_auc_score(y_true, y_score, sample_weight=None):
    267         if len(np.unique(y_true)) != 2:
--> 268             raise ValueError("Only one class present in y_true. ROC AUC score "
    269                              "is not defined in that case.")
    270 

ValueError: Only one class present in y_true. ROC AUC score is not defined in that case.

我的环境：

python==3.7.2

sklearn==0.19.2

我的问题：

这是一个bug，还是我在错误地使用它？

- Emmanuel-Lin

Answer 1

默认情况下，scikit-learn的交叉验证功能会带来一个不必要的麻烦：数据不会被随机洗牌。实际上，将数据进行洗牌是个好主意，因此最好将洗牌作为默认选择。当然，这也意味着在首次使用cross_val_score时需要提供一个洗牌参数，但不幸的是，官方文档中并没有提到该参数。

以下是发生的情况：iris数据集的150个样本被分层抽样：

iris.target[0:50]
# result
array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0])

iris.target[50:100]
# result:
array([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1])

iris.target[100:150]
# result:
array([2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
       2, 2, 2, 2, 2, 2])

现在，使用上述展示的分层方式，150个样本进行三折交叉验证程序，并出现错误消息：

ValueError: Only one class present in y_true

希望以下内容更加易懂：在三个验证集中，每一个只有一个标签存在，因此无法计算ROC（更不用说在每个验证集中模型看到的标签是训练集中未见过的）。 因此，请在之前打乱您的数据：

from sklearn.utils import shuffle
X_s, y_s = shuffle(X, y)
cross_val_score(model, X_s, y_s, cv=3, scoring="roc_auc")

你只需要按照要求进行操作，就会没问题了。

- desertnaut

---

太棒了，它像魔法一样运行。我错误地认为交叉验证会进行 suffling。谢谢。 - Emmanuel-Lin