获取ValueError:对LogisticRegression应用拟合方法时，索引数组的元素数少于nnz

Question

我正在尝试训练一个在onevsrest分类器中嵌套LogisticRegression的模型。当对分类器应用fit方法时，它抛出ValueError: indices array has fewer than nnz elements.

代码：

params = {'estimator__C' : [1]}
classifier_2 = OneVsRestClassifier(LogisticRegression(penalty='l2',solver = 'lbfgs'), n_jobs=-1)
grid_search = GridSearchCV(classifier_2,params,n_jobs = -1,cv = 2,verbose = 50,scoring = 'f1_micro')
grid_search.fit(x_train_multilabel, y_train)

+以下是指数和CSR矩阵的形状。你知道吗

print(x_train_multilabel.shape)
print(len(x_train_multilabel.indices))

输出

(400000, 83425)
9653403