I am playing around with some data from a Kaggle competition on text_analysis, and I keep getting this rather weird error described in the title whenever I try to fit my algorithm. I looked it up, and it had something to with my matrix being to densely populated with nonzero elements while presented as a sparse matrix. I reckon this problem lies with my train_labels below in the code, the labels consist of 24 columns which isn't very common to begin with, labels are floats between 0 and 1 (including 0 and 1). Despite having some idea on what the problem is, I have no idea on how to tackle it properly, and my previous tries haven't worked out so well. Do you guys have any suggestions on how I could solve this?
Code:
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")
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
I suspect that my labels are the problem, so here are the labels:
In [12]:
undefined
import pandas as pd
import numpy as np
import os
dir = "C:\Users\Anonymous\Desktop\KAGA FOLDER\Hashtags"
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)
