svm.grid_search_cv(n_jobs=4, verbose=5)
test_error = svm.get_test_error()
print "Test error in held out set: " + str(test_error)
print "=" * 20
svm = SVMClassifier(train_reviews, train_labels, ngram_range=(1,2),
test_texts=test_reviews, test_labels=test_labels,
compute_features=True)
svm.grid_search_cv(n_jobs=4, verbose=5)
test_error = svm.get_test_error()
print "Test error in held out set: " + str(test_error)
print "=" * 20
# Simple bag of words with a logistic classifier
lr = LogisticClassifier(train_reviews, train_labels,
test_texts=test_reviews, test_labels=test_labels,
compute_features=True)
lr.grid_search_cv(verbose=5, n_jobs=4)
test_error = lr.get_test_error()
print "Test error in held out set: " + str(test_error)
print "=" * 20
lr = LogisticClassifier(train_reviews, train_labels, ngram_range=(1,2),
test_texts=test_reviews, test_labels=test_labels,
compute_features=True)
lr.grid_search_cv(verbose=5, n_jobs=4)
test_error = lr.get_test_error()
print "Test error in held out set: " + str(test_error)
print "=" * 20
# SGD up to 3-grams
print "Test error: " + str(test_error)
print "SGD Classifier"
sgd = SGDTextClassifier(train_reviews,
train_labels,
test_texts=test_reviews,
test_labels=test_labels)
#train_error = sgd.get_training_error()
#test_error = sgd.get_test_error()
#print "Training error: " + str(train_error)
#print "Test error: " + str(test_error)
sgd.set_bag_of_ngrams()
sgd.grid_search_cv(verbose=0, n_jobs=4)
print "Logistic classifier"
sgd = LogisticClassifier()
sgd.set_training_data(train_reviews, train_labels)
sgd.set_test_data(test_reviews, test_labels)
sgd.set_bag_of_ngrams()
sgd.train()
train_error = sgd.get_training_error()
test_error = sgd.get_test_error()
print "Training error: " + str(train_error)
print "Test error: " + str(test_error)
print "SVM classifier"
sgd = SVMClassifier()
sgd.set_training_data(train_reviews, train_labels)
sgd.set_test_data(test_reviews, test_labels)
sgd.set_bag_of_ngrams()
svm = SVMClassifier(train_reviews,
train_labels,
ngram_range=(1, 2),
test_texts=test_reviews,
test_labels=test_labels,
compute_features=True)
svm.grid_search_cv(n_jobs=4, verbose=5)
test_error = svm.get_test_error()
print "Test error in held out set: " + str(test_error)
print "=" * 20
# Simple bag of words with a logistic classifier
lr = LogisticClassifier(train_reviews,
train_labels,
test_texts=test_reviews,
test_labels=test_labels,
compute_features=True)
lr.grid_search_cv(verbose=5, n_jobs=4)
test_error = lr.get_test_error()
print "Test error in held out set: " + str(test_error)
print "=" * 20
lr = LogisticClassifier(train_reviews,
train_labels,
ngram_range=(1, 2),
test_texts=test_reviews,
test_labels=test_labels,
compute_features=True)
lr.grid_search_cv(verbose=5, n_jobs=4)
test_error = lr.get_test_error()
nb.set_bag_of_ngrams() # Also can compute bag of words manually
nb.grid_search_cv(n_jobs=4)
# Now shit with bigrams too
sgd = SGDTextClassifier(train_reviews, train_labels, ngram_range=(1,2),
test_texts=test_reviews, test_labels=test_labels,
compute_features=True)
sgd.grid_search_cv(n_jobs=4, verbose=1)
nb = NaiveBayesClassifier(train_reviews, train_labels, ngram_range=(1,2),
test_texts=test_reviews, test_labels=test_labels,
compute_features=True)
nb.grid_search_cv(n_jobs=4, verbose=1)
lr = LogisticClassifier(train_reviews, train_labels, ngram_range=(1,2),
test_texts=test_reviews, test_labels=test_labels,
compute_features=True)
lr.grid_search_cv(verbose=5, n_jobs=4)
# print "Naive Bayes"
# nb = NaiveBayesClassifier()
# nb.set_training_data(train_reviews, train_labels)
# nb.set_test_data(test_reviews, test_labels)
# nb.set_bag_of_ngrams()
#
# nb.train()
# train_error = nb.get_training_error()
# test_error = nb.get_test_error()
# print "Training error: " + str(train_error)
# print "Test error: " + str(test_error)