Skip to main content
 首页 » 编程设计

python之Sklearn LogisticRegressionCV 的类似数组的输入

2025年02月15日16bjzhanghao

最初,我从 .csv 中读取数据文件,但在这里我从列表构建数据框,以便可以重现问题。目的是使用 LogisticRegressionCV 训练具有交叉验证的逻辑回归模型。 .

indeps = ['M', 'F', 'M', 'F', 'M', 'M', 'F', 'M', 'M', 'F', 'F', 'F', 'F', 'F', 'M', 'F', 'F', 'F', 'F', 'F', 'M', 'F', 'F', 'M', 'M', 'F', 'F', 'F', 'M', 'F', 'F', 'F', 'M', 'F', 'M', 'F', 'F', 'F', 'M', 'M', 'M', 'F', 'M', 'M', 'M', 'F', 'M', 'M', 'F', 'F'] 
dep = [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 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] 
 
data = [indeps, dep]  
cols = ['state', 'cat_bins'] 
 
data_dict = dict((x[0], x[1]) for x in zip(cols, data)) 
 
df = pd.DataFrame.from_dict(data_dict) 
df.tail() 
 
    cat_bins    state 
45  0.0           F 
46  0.0           M 
47  0.0           M 
48  0.0           F 
49  0.0           F 
 
 
'''Use Pandas' to encode independent variables.  Notice that 
 we are returning a sparse dataframe ''' 
 
def heat_it2(dataframe, lst_of_columns): 
    dataframe_hot = pd.get_dummies(dataframe, 
                                   prefix = lst_of_columns, 
                                   columns = lst_of_columns, sparse=True,) 
    return dataframe_hot 
 
train_set_hot = heat_it2(df, ['state']) 
train_set_hot.head(2) 
 
    cat_bins    state_F     state_M 
0     1.0         0            1 
1     1.0         1            0 
 
'''Use the dataframe to set up the prospective inputs to the model as numpy arrays''' 
 
indeps_hot = ['state_F', 'state_M'] 
 
X = train_set_hot[indeps_hot].values 
y = train_set_hot['cat_bins'].values 
 
print 'X-type:', X.shape, type(X) 
print 'y-type:', y.shape, type(y) 
print 'X has shape, is an array and has length:\n', hasattr(X, 'shape'), hasattr(X, '__array__'), hasattr(X, '__len__') 
print 'yhas shape, is an array and has length:\n', hasattr(y, 'shape'), hasattr(y, '__array__'), hasattr(y, '__len__') 
print 'X does have attribute fit:\n',hasattr(X, 'fit') 
print 'y does have attribute fit:\n',hasattr(y, 'fit') 
 
X-type: (50, 2) <type 'numpy.ndarray'> 
y-type: (50,) <type 'numpy.ndarray'> 
X has shape, is an array and has length: 
True True True 
yhas shape, is an array and has length: 
True True True 
X does have attribute fit: 
False 
y does have attribute fit: 
False 

因此,回归器的输入似乎具有 .fit 的必要属性方法。它们是 具有正确形状的 numpy 数组。 X是维度为 [n_samples, n_features] 的数组, 和 y是一个形状为 [n_samples,] 的向量这是文档:

fit(X, y, sample_weight=None)[source]

Fit the model according to the given training data. 
Parameters:  
 
X : {array-like, sparse matrix}, shape (n_samples, n_features) 
 
    Training vector, where n_samples is the number of samples and n_features is the number of features. 
  y : array-like, shape (n_samples,) 
 
Target vector relative to X. 

....

现在我们尝试拟合回归量:

logmodel = LogisticRegressionCV(Cs =1, dual=False , scoring = accuracy_score, penalty = 'l2') 
logmodel.fit(X, y) 
 
... 
 
    TypeError: Expected sequence or array-like, got estimator LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True, 
    intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1, 
    penalty='l2', random_state=None, solver='liblinear', tol=0.0001, 
    verbose=0, warm_start=False) 

错误消息的来源似乎在 scikits 的 validation.py 模块中,here .

引发此错误消息的唯一代码部分是以下函数片段:

def _num_samples(x): 
    """Return number of samples in array-like x.""" 
    if hasattr(x, 'fit'): 
        # Don't get num_samples from an ensembles length! 
        raise TypeError('Expected sequence or array-like, got ' 
                        'estimator %s' % x) 
    etc. 

问题:由于我们用来拟合模型的参数(X 和 y)不具有“fit”属性,为什么会出现此错误消息

在带有 scikit-learn 18.01-3 和 pandas 0.19.2-2 的 Canopy 1.7.4.3348(64 位)上使用 python 2.7

谢谢你的帮助:)

请您参考如下方法:

问题似乎出在 scoring 参数中。您已通过 accuracy_score。 accuracy_score 的签名是 accuracy_score(y_true, y_pred[, ...])。但是在模块logistic.py

if isinstance(scoring, six.string_types): 
    scoring = SCORERS[scoring] 
for w in coefs: 
    // Other code 
    if scoring is None: 
        scores.append(log_reg.score(X_test, y_test)) 
    else: 
        scores.append(scoring(log_reg, X_test, y_test)) 

因为你已经通过了 accuracy_score,所以它不符合上面的第一行。 并且 scores.append(scoring(log_reg, X_test, y_test)) 用于对估计器进行评分。但正如我上面所说,这里的参数与 accuracy_score 所需的参数不匹配。因此错误。

解决方法:使用 make_scorer LogisticRegressionCV 中的 (accuracy_score) 用于评分或简单地传递字符串 'accuracy'

logmodel = LogisticRegressionCV(Cs =1, dual=False ,  
                                scoring = make_scorer(accuracy_score),  
                                penalty = 'l2') 
 
                         OR 
 
logmodel = LogisticRegressionCV(Cs =1, dual=False ,  
                                scoring = 'accuracy',  
                                penalty = 'l2') 

注意:

这可能是 logistic.py 模块的一部分或 LogisticRegressionCV 文档中的错误,他们应该已经阐明了评分函数的签名。

您可以提交 an issue to the github and see how it goes 完成