最初,我从.csv
文件中读取数据,但是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
训练逻辑回归模型并进行交叉验证。
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,回归量的输入似乎具有.fit
方法的必要属性。它们是 numpy数组,形状正确。 X
是一个维度为[n_samples, n_features]
的数组,y
是一个形状为[n_samples,]
的向量。这是文档:
适合(X,y,sample_weight =无)[来源]
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.
...
现在我们尝试适应回归量:
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是在scikits的validation.py模块here中。
引发此错误消息的唯一代码部分是以下function-snippet:
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:由于我们拟合模型的参数(X
和y
)没有属性“fit”,为什么会出现此错误消息
在Canopy 1.7.4.3348(64位)上使用python 2.7,使用scikit-learn 18.01-3和pandas 0.19.2-2
感谢您的帮助:)
答案 0 :(得分:1)
问题似乎出现在scoring
参数中。您已通过accuracy_score
。 accuracy_score
的签名是accuracy_score(y_true, y_pred[, ...])
。但在模块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))
由于您已通过accuracy_score
,因此它不适合上面的第一行。
scores.append(scoring(log_reg, X_test, y_test))
用于评估估算工具。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LogisticRegressionCV中的make_scorer(accuracy_score)进行评分,或者只是传递字符串'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')
注意强>:
这可能是部分logistic.py
模块或LogisticRegressionCV文档中的错误,他们应该澄清评分函数的签名。