值错误:不允许使用负尺寸

Question

我正在text_analysis上玩一些来自Kaggle比赛的数据，每当我尝试适应我的算法时，我都会在标题中描述这个相当奇怪的错误。我查了一下，它与我的矩阵有一些关系，即在呈现为稀疏矩阵的同时密集填充非零元素。我认为这个问题出train_labels在下面的代码中，标签由 24 列组成，这在开始时不是很常见，标签是 0 到 1（包括 0 和 1）之间的浮点数。尽管对问题是什么有一些想法，但我不知道如何正确解决它，而且我以前的尝试效果不佳。你们对我如何解决这个问题有什么建议吗？

法典：

import numpy as np
import pandas as p
import nltk
from sklearn.feature_extraction.text import TfidfVectorizer
import os
from sklearn.linear_model  import RidgeCV
dir = "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 alist
def 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")

追踪：

Traceback (most recent call last):
  File "C:UsersAnonymousworkspacefinal_submissionsrclinearSVM.py", line 56, in <module>
    ridge_classifier.fit(train,train_labels)
  File "C:Python27libsite-packagessklearnlinear_modelridge.py", line 817, in fit
    estimator.fit(X, y, sample_weight=sample_weight)
  File "C:Python27libsite-packagessklearnlinear_modelridge.py", line 724, in fit
    v, Q, QT_y = _pre_compute(X, y)
  File "C:Python27libsite-packagessklearnlinear_modelridge.py", line 609, in _pre_compute
    K = safe_sparse_dot(X, X.T, dense_output=True)
  File "C:Python27libsite-packagessklearnutilsextmath.py", line 78, in safe_sparse_dot
    ret = a * b
  File "C:Python27libsite-packagesscipysparsebase.py", line 303, in __mul__
    return self._mul_sparse_matrix(other)
  File "C:Python27libsite-packagesscipysparsecompressed.py", line 520, in _mul_sparse_matrix
    indices = np.empty(nnz, dtype=np.intc)
ValueError: negative dimensions are not allowed

我怀疑我的标签有问题，所以这里是标签：

In [12]:
undefined

import pandas as pd
import numpy as np
import os
dir = "C:UsersAnonymousDesktopKAGA FOLDERHashtags"
labels = np.array(pd.read_csv(os.path.join(dir,"train.csv")))[:,4:]
labels

Out[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]:
undefined

labels.shape
Out[13]:
(77946L, 24L)

Answer 1

问题是由于大小不匹配。

train_labels实际上是所有数据的类。train和train_labels的大小应匹配。