我在玩一个关于文本分析的Kaggle竞赛的数据时，尝试拟合我的算法时总是遇到标题中描述的奇怪错误。我查了一下，发现这与我的矩阵在表示为稀疏矩阵时非零元素过于密集有关。我认为问题出在代码中的train_labels上，我的标签有24列，这本身就不太常见，标签是介于0和1之间的浮点数（包括0和1）。尽管我对问题有些了解，但我不知道如何正确解决它，我之前的尝试效果也不太好。你们有什么建议可以解决这个问题吗？

代码:

import numpy as npimport pandas as pimport nltkfrom sklearn.feature_extraction.text import TfidfVectorizerimport osfrom sklearn.linear_model  import RidgeCVdir = "C:/Users/Anonymous/Desktop/KAGA FOLDER/Hashtags"def clean_the_text(data):    alist = []    data = nltk.word_tokenize(data)    for j in data:        alist.append(j.rstrip('\n'))    alist = " ".join(alist)    return alistdef loop_data(data):    for i in range(len(data)):        data[i] = clean_the_text(data[i])    return data      if __name__ == "__main__":    print("loading data")    train_text = loop_data(list(np.array(p.read_csv(os.path.join(dir,"train.csv")))[:,1]))    test_set = loop_data(list(np.array(p.read_csv(os.path.join(dir,"test.csv")))[:,1]))    train_labels  = np.array(p.read_csv(os.path.join(dir,"train.csv")))[:,4:]    #Vectorizing    vectorizer = TfidfVectorizer(max_features = 10000,strip_accents = "unicode",analyzer = "word")    ridge_classifier = RidgeCV(alphas = [0.001,0.01,0.1,1,10])    all_data = train_text + test_set    train_length  = len(train_text)    print("fitting Vectorizer")    vectorizer.fit(all_data)    print("transforming text")    all_data = vectorizer.transform(all_data)    train = all_data[:train_length]    test = all_data[train_length:]    print("fitting and selecting models")     ridge_classifier.fit(train,train_labels)    print("predicting")    pred = ridge_classifier.predict(test)    np.savetxt(dir +"submission.csv", pred, fmt = "%d", delimiter = ",")    print("submission_file created")

Traceback:

Traceback (most recent call last):  File "C:\Users\Anonymous\workspace\final_submission\src\linearSVM.py", line 56, in <module>    ridge_classifier.fit(train,train_labels)  File "C:\Python27\lib\site-packages\sklearn\linear_model\ridge.py", line 817, in fit    estimator.fit(X, y, sample_weight=sample_weight)  File "C:\Python27\lib\site-packages\sklearn\linear_model\ridge.py", line 724, in fit    v, Q, QT_y = _pre_compute(X, y)  File "C:\Python27\lib\site-packages\sklearn\linear_model\ridge.py", line 609, in _pre_compute    K = safe_sparse_dot(X, X.T, dense_output=True)  File "C:\Python27\lib\site-packages\sklearn\utils\extmath.py", line 78, in safe_sparse_dot    ret = a * b  File "C:\Python27\lib\site-packages\scipy\sparse\base.py", line 303, in __mul__    return self._mul_sparse_matrix(other)  File "C:\Python27\lib\site-packages\scipy\sparse\compressed.py", line 520, in _mul_sparse_matrix    indices = np.empty(nnz, dtype=np.intc)ValueError: negative dimensions are not allowed

我怀疑我的标签是问题所在，所以这里是我的标签:

In [12]:undefinedimport pandas as pdimport numpy as npimport osdir = "C:\Users\Anonymous\Desktop\KAGA FOLDER\Hashtags"labels = np.array(pd.read_csv(os.path.join(dir,"train.csv")))[:,4:]labelsOut[12]:array([[0.0, 0.0, 1.0, ..., 0.0, 0.0, 0.0],       [0.0, 0.0, 0.0, ..., 0.0, 0.0, 0.0],       [0.0, 0.0, 0.0, ..., 0.0, 0.0, 0.0],       ...,        [0.0, 0.0, 0.0, ..., 1.0, 0.0, 0.0],       [0.0, 0.385, 0.41, ..., 0.0, 0.0, 0.0],       [0.0, 0.20199999999999999, 0.395, ..., 0.0, 0.0, 0.0]], dtype=object)In [13]:undefinedlabels.shapeOut[13]:(77946L, 24L)

回答：

问题是由于尺寸不匹配造成的。

train_labels实际上是所有数据的类别。 train 和 train_labels 的尺寸应该匹配。