使用随机森林时，scikit 中的“ValueError: max_features must be in (0, n_features]”

Posted 2023-03-05

【中文标题】使用随机森林时，scikit 中的“ValueError: max_features must be in (0, n_features]”

【英文标题】："ValueError: max_features must be in (0, n_features] " in scikit when using random forest

【发布时间】：2017-06-23 16:28:05

【问题描述】：

我有一个包含 20 个特征和 840 行的数据集。 我已经优化了分类器（随机森林）。我的参数是 n_estimators=100 和 max_features=5。 我想对每个特征进行分类。我的意思是我想知道每个特征的预测准确性。但是当我使用我的代码时，我得到了一个错误。 我正在使用 scikit 版本。 18.

我该如何解决这个问题？

for name in ["AWA"]: 
    x=sio.loadmat('/home/TrainVal/_Fp1.mat'.format(name))['x'] 
    s_y=sio.loadmat('/home/TrainVal/_Fp1.mat'.format(name))['y']
    y=np.ravel(s_y)

    print(name, x.shape, y.shape) 
    print("")


    clf = make_pipeline(preprocessing.RobustScaler(), RandomForestClassifier(n_estimators = 100, 
                                                                         max_features=5, n_jobs=-1)) 
    #########10x10 SSS##############
    print("10x10")

    for i in range(x.shape[1]): 
        xA=x[:, i].reshape(-1,1)

        xSSSmean = [] 
        for j in range(10):
            sss = StratifiedShuffleSplit(n_splits=10, test_size=0.1, random_state=j) 
            scoresSSS = cross_val_score(clf, xA, y, cv=sss)
            xSSSmean.append(scoresSSS.mean()) 

        result_list.append(np.mean(xSSSmean))  
        plt.bar(i, np.mean(xSSSmean)*100, align = 'center')      
        plt.ylabel('Accuracy')
        plt.xlabel('Features')    
        plt.title('Accuracy per feature: _RF_Fp1(20)'.format(name)) 

        xticks=np.arange(i+1)
        plt.xticks(xticks, rotation = 'vertical')
    plt.show()





#THE ERROR


ValueError                                Traceback (most recent call last)
<ipython-input-2-a5faae7f83a2> in <module>()
     24 
     25             sss = StratifiedShuffleSplit(n_splits=10, test_size=0.1, random_state=j)#ver18
---> 26             scoresSSS = cross_val_score(clf, xA, y, cv=sss)
     27             xSSSmean.append(scoresSSS.mean()) 
     28             #print(scoresSSS)

 /home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/model_selection/_validation.py in cross_val_score(estimator, X, y, groups, scoring, cv, n_jobs, verbose, fit_params, pre_dispatch)
    138                                               train, test, verbose, None,
    139                                               fit_params)
--> 140                       for train, test in cv_iter)
    141     return np.array(scores)[:, 0]
    142 

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/externals/joblib/parallel.py in __call__(self, iterable)
    756             # was dispatched. In particular this covers the edge
    757             # case of Parallel used with an exhausted iterator.
--> 758             while self.dispatch_one_batch(iterator):
    759                 self._iterating = True
    760             else:

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/externals/joblib/parallel.py in dispatch_one_batch(self, iterator)
    606                 return False
    607             else:
--> 608                 self._dispatch(tasks)
    609                 return True
    610 

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/externals/joblib/parallel.py in _dispatch(self, batch)
    569         dispatch_timestamp = time.time()
    570         cb = BatchCompletionCallBack(dispatch_timestamp, len(batch), self)
--> 571         job = self._backend.apply_async(batch, callback=cb)
    572         self._jobs.append(job)
    573 

 /home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/externals/joblib/_parallel_backends.py in apply_async(self, func, callback)
    107     def apply_async(self, func, callback=None):
    108         """Schedule a func to be run"""
--> 109         result = ImmediateResult(func)
    110         if callback:
    111             callback(result)

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/externals/joblib/_parallel_backends.py in __init__(self, batch)
    324         # Don't delay the application, to avoid keeping the input
    325         # arguments in memory
--> 326         self.results = batch()
    327 
    328     def get(self):

/home/karim/anaconda2/envs/scikit18/lib/python3.5/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):

 /home/karim/anaconda2/envs/scikit18/lib/python3.5/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):

/home/karim/anaconda2/envs/scikit18/lib/python3.5/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)
    236             estimator.fit(X_train, **fit_params)
    237         else:
--> 238             estimator.fit(X_train, y_train, **fit_params)
    239 
    240     except Exception as e:

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/pipeline.py in fit(self, X, y, **fit_params)
    268         Xt, fit_params = self._fit(X, y, **fit_params)
    269         if self._final_estimator is not None:
--> 270             self._final_estimator.fit(Xt, y, **fit_params)
    271         return self
    272 

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/ensemble/forest.py in fit(self, X, y, sample_weight)
    324                     t, self, X, y, sample_weight, i, len(trees),
    325                     verbose=self.verbose, class_weight=self.class_weight)
--> 326                 for i, t in enumerate(trees))
    327 
    328             # Collect newly grown trees

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/externals/joblib/parallel.py in __call__(self, iterable)
    756             # was dispatched. In particular this covers the edge
    757             # case of Parallel used with an exhausted iterator.
--> 758             while self.dispatch_one_batch(iterator):
    759                 self._iterating = True
    760             else:

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/externals/joblib/parallel.py in dispatch_one_batch(self, iterator)
    606                 return False
    607             else:
--> 608                 self._dispatch(tasks)
    609                 return True
    610 

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/externals/joblib/parallel.py in _dispatch(self, batch)
    569         dispatch_timestamp = time.time()
    570         cb = BatchCompletionCallBack(dispatch_timestamp, len(batch), self)
--> 571         job = self._backend.apply_async(batch, callback=cb)
    572         self._jobs.append(job)
    573 

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/externals/joblib/_parallel_backends.py in apply_async(self, func, callback)
    107     def apply_async(self, func, callback=None):
    108         """Schedule a func to be run"""
--> 109         result = ImmediateResult(func)
    110         if callback:
    111             callback(result)

 /home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/externals/joblib/_parallel_backends.py in __init__(self, batch)
    324         # Don't delay the application, to avoid keeping the input
    325         # arguments in memory
--> 326         self.results = batch()
    327 
    328     def get(self):

/home/karim/anaconda2/envs/scikit18/lib/python3.5/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):

 /home/karim/anaconda2/envs/scikit18/lib/python3.5/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):

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/ensemble/forest.py in _parallel_build_trees(tree, forest, X, y, sample_weight, tree_idx, n_trees, verbose, class_weight)
    118             curr_sample_weight *= compute_sample_weight('balanced', y, indices)
    119 
--> 120         tree.fit(X, y, sample_weight=curr_sample_weight, check_input=False)
    121     else:
    122         tree.fit(X, y, sample_weight=sample_weight, check_input=False)

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/tree/tree.py in fit(self, X, y, sample_weight, check_input, X_idx_sorted)
    737             sample_weight=sample_weight,
    738             check_input=check_input,
--> 739             X_idx_sorted=X_idx_sorted)
    740         return self
    741 

/home/karim/anaconda2/envs/scikit18/lib/python3.5/site-packages/sklearn/tree/tree.py in fit(self, X, y, sample_weight, check_input, X_idx_sorted)
    244             raise ValueError("max_depth must be greater than zero. ")
    245         if not (0 < max_features <= self.n_features_):
--> 246             raise ValueError("max_features must be in (0, n_features]")
    247         if not isinstance(max_leaf_nodes, (numbers.Integral, np.integer)):
    248             raise ValueError("max_leaf_nodes must be integral number but was "

ValueError: max_features must be in (0, n_features]

【参考方案1】：

所以我设法解决了这个问题！！！ :) 在scikit page 中说：

*如果是浮点数，那么 max_features 是一个百分比，并且在每次拆分时都会考虑 int(max_features * n_features) 个特征。*

我的价值：

列表项

n_features=20。这是在 int 中。这是我在数据集中拥有的特征数量。

max_features：这是我想要使用的功能数量。但是它们在 int 中，所以我必须将它们变成 float

要将它变成 float 我必须使用 scikit 中的公式：

int(max_features * n_features)
int(x * 20)=2
x=0.1

我们必须假设我只想使用 20 个功能中的 2 个。

x 是 float

中的百分比

我将 max_features 中的值从 int 更改为 float。就像这样：

最大特征：

(int) (浮点数)

20 = 1.0

15 = 0.75

10 = 0.5

5 = 0.25

2 = 0.1

示例

#Instead of: 
clf = make_pipeline(preprocessing.RobustScaler(), RandomForestClassifier(n_estimators = 100, 
                   max_features=5, n_jobs=-1)) 

#I did:
clf = make_pipeline(preprocessing.RobustScaler(), RandomForestClassifier(n_estimators = 100, 
                   max_features=0.25, n_jobs=-1))

【讨论】：

20 = 1.0 怎么算？

嗨。在 scikit 的网页中，您可以看到：“如果浮动，则 max_features 是一个百分比，并且在每次拆分时都会考虑 int(max_features * n_features) 个特征。我将在我的答案中进行解释。