Question

我正在构建一个程序，该程序将多个标签/标签分配给文本描述。我正在使用OneVsRestClassifier标记我的文字描述。 xTrain，xTest和yTrain均为'numpy.ndarray'。考虑到我已经以正确的方式拆分了训练和测试数据，这似乎有些奇怪。下面是我的代码：

xTrain, xTest, yTrain, yTest = train_test_split(x, y, test_size=0.2)

nb_clf = MultinomialNB()
sgd = SGDClassifier()
lr = LogisticRegression()
mn = MultinomialNB()

print("xTrain.shape = " + str(xTrain.shape))
print("xTest.shape = " + str(xTest.shape))
print("yTrain.shape = " + str(yTrain.shape))
print("yTest.shape = " + str(yTest.shape))

print("type(xTrain) = " + str(type(xTrain)))
print("type(xTest) = " + str(type(xTest)))

xTrain = csr_matrix(xTrain).toarray()
xTest = csr_matrix(xTest).toarray()
yTrain = csr_matrix(yTrain).toarray()

print("type(xTrain) = " + str(type(xTrain)))

for classifier in [nb_clf, sgd, lr, mn]:
    clf = OneVsRestClassifier(classifier)
    clf.fit(xTrain.astype("U"), yTrain.astype("U"))
    y_pred = clf.predict(xTest)
    print("\ny_pred:")
    print(y_pred)

x输出：

  (1466, 1292)  0.13531037414782607
  (1466, 1238)  0.21029405543816293
  (1466, 988)   0.04688335706505732
  ...
  ...

您的输出：

[[0 0 0 ... 1 0 0]
 [0 0 0 ... 0 0 0]
 [0 0 0 ... 1 0 0]
 ...
 [0 0 0 ... 0 0 0]
 [0 0 0 ... 0 0 0]
 [0 0 0 ... 0 0 0]]

打印语句输出：

xTrain.shape = (1173, 13817)
xTest.shape = (294, 13817)
yTrain.shape = (1173, 28)
yTest.shape = (294, 28)
type(xTrain) = <class 'scipy.sparse.csr.csr_matrix'>
type(xTest) = <class 'scipy.sparse.csr.csr_matrix'>
type(xTrain) = <class 'numpy.ndarray'>
type(xTest) = <class 'numpy.ndarray'>
type(yTrain) = <class 'numpy.ndarray'>

错误（在clf.fit行）：

ValueError：标签不支持多输出目标数据二值化

Answer 1

请首先在程序中阐明特征尺寸和样本量。对于目标功能（y），标签不应该被一键编码。例如，应该是[3]而不是[0 0 0 1]

Sklearn Multilabel ML：ValueError：标签二值化不支持多输出目标数据

1 个答案: