def score(classifier):
classifier = nltk.SklearnClassifier(classifier) # 在nltk 中使用scikit-learn 的接口
classifier.train(train) # 训练分类器
classifier = joblib.load('model.m')
# joblib.dump(classifier,'model.m')
pred = classifier.classify_many(dev) # 对开发测试集的数据进行分类，给出预测的标签
return precision_recall_fscore_support(tag_dev, pred)
def get_classifier(classifier,
train,
mealgorithm='GIS',
sklalgorithm='bernoulli'):
if classifier == 'NaiveBayes':
return nltk.NaiveBayesClassifier.train(train)
elif classifier == 'DecisionTree':
return nltk.DecisionTreeClassifier.train(train)
elif classifier == 'Maxent':
return nltk.DecisionTreeClassifier.train(train, mealgorithm)
elif classifier == 'Sklearn':
if sklalgorithm == 'bernoulli':
return nltk.SklearnClassifier(BernoulliNB()).train(train)
elif sklalgorithm == 'svc':
return nltk.SklearnClassifier(SVC(), sparse=False).train(train)
elif classifier == 'SVM':
pass
def score2(classifier):
classifier = nltk.SklearnClassifier(classifier) # 在nltk 中使用scikit-learn 的接口
classifier.train(train) # 训练分类器
pred = classifier.classify_many(dev) # 对开发测试集的数据进行分类，给出预测的标签
# print(classification_report(tag_dev, pred))
# print('\n')
return accuracy_score(tag_dev, pred) # 对比分类预测结果和人工标注的正确结果，给出分类器准确度
def score(classifier):
classifier = nltk.SklearnClassifier(classifier) # 在nltk 中使用scikit-learn的接口
classifier.train(train) # 训练分类器
pred = classifier.classify_many(dev) # 对开发测试集的数据进行分类，给出预测的标签
a_s = accuracy_score(tag_dev, pred)
p_s = precision_score(tag_dev, pred, average='binary', pos_label="pos")
r_s = recall_score(tag_dev, pred, average='binary', pos_label="pos")
f1_s = f1_score(tag_dev, pred, average='binary', pos_label="pos")
return a_s, p_s, r_s, f1_s # 对比分类预测结果和人工标注的正确结果，给出分类器准确度
def score(classifier):
classifier = nltk.SklearnClassifier(classifier)
classifier.train(train)
joblib.dump(classifier, 'LogisticRegression.model')
pred = classifier.classify_many(data)
n = 0
s = len(pred)
for i in range(0, s):
if pred[i] == tag[i]:
n = n + 1
return n / s
def saveBestClassifier(classifier,
fileName,
featureSet,
testRatio,
sklean=False):
rows = len(featureSet)
print("Training %s 100 times..." % (fileName))
if sklean:
classifierBest = nltk.SklearnClassifier(classifier)
else:
classifierBest = classifier
maxAcc = 0
for i in range(100):
random.shuffle(featureSet)
# Spliting into train and test set
trainingSet = featureSet[:1900]
testingSet = featureSet[1900:]
if sklean:
classifierCurr = nltk.SklearnClassifier(classifier).train(
trainingSet)
else:
classifierCurr = classifier.train(trainingSet)
currAcc = nltk.classify.accuracy(classifierCurr, testingSet)
if currAcc > maxAcc:
classifierBest = classifierCurr
maxAcc = currAcc
print("New best accuracy for %s:" % (fileName), maxAcc)
print("Done: %.2f" % (i / 100))
print("Saving %s with max accuracy:" % (fileName), maxAcc)
file = open("./classifiers/" + fileName + ".pickle", "wb")
pickle.dump(classifierBest, file)
file.close()
def classify(self, training_set, classifierDumpFile):
"""
:param training_set: Features to train the classifier
:param classifierDumpFile: File name to dump -save- the trained data.
:return: classifier
"""
# Train the module of the specified algorithm using the created training set.
self.classifier = nltk.SklearnClassifier(LinearSVC())
self.classifier.train(training_set)
#self.classifier = nltk.NaiveBayesClassifier.train(training_set)
# Save the trainied data in pickle for further use.
self.dump_files(self.classifier, self.classifierDumpFile)
return self.classifier
def score(classifier):
classifier = nltk.SklearnClassifier(classifier) #在nltk 中使用scikit-learn 的接口
classifier.train(train) #训练分类器
pred = classifier.classify_many(data) #对开发测试集的数据进行分类，给出预测的标签
return accuracy_score(tag, pred)
# naive_bayes = NaiveBayesClassifier.train(train_set)
# print("Accuracy - Naive Bayes Classifier: ")
# print(nltk.classify.accuracy(naive_bayes, test_set))
# print("Most informative features - Naive Bayes Classifier:")
# print(naive_bayes.show_most_informative_features())
# maxent = MaxentClassifier.train(train_set, 'GIS', trace=0,
# encoding=None, gaussian_prior_sigma=0, max_iter=100)
# print("Accuracy - Max Entropy Classifier: ")
# print(nltk.classify.accuracy(maxent, test_set))
# print("Most informative features - Max Entropy Classifier:")
# print(maxent.show_most_informative_features())
linear_svm_classifier = nltk.SklearnClassifier(LinearSVC(C=2.0, dual=True, fit_intercept=True,
intercept_scaling=0.1, loss='squared_hinge',
max_iter=1500, penalty='l2', random_state=0,
tol=0.0001), sparse=False)
linear_svm_classifier.train(train_set)
print("Accuracy - Linear SVM Classifier: ")
print(nltk.classify.accuracy(linear_svm_classifier, test_set))
nonlinear_svm = SklearnClassifier(SVC(gamma='scale', kernel='poly', coef0 = 5.0, degree = 5, C = 5.0, shrinking=True, probability=False, tol=1e-3), sparse=False).train(train_set)
print("Accuracy - Nonlinear SVM: ")
print(nltk.classify.accuracy(nonlinear_svm, test_set))
random_forest = SklearnClassifier(RandomForestClassifier(n_estimators = 100,
criterion = 'gini',
max_depth = 5,
min_samples_split = 2,
all_words = nltk.FreqDist(all_words)
# print(all_words.most_common(15))
word_features = list(all_words.keys())[:3000]
def find_features(document):
words = nltk.word_tokenize(document)
features = {}
for w in word_features:
features[w] = (w in words)
return features
featuresets = [(find_features(rev), category) for (category, rev) in documents]
random.shuffle(featuresets)
training_set = featuresets[:4500]
testing_set = featuresets[4500:]
# classifier_f = open("Wilfred_naivebayes.pickle", "rb")
# classifier = pickle.load(classifier_f)
# classifier_f.close()
#
# classifier = nltk.NaiveBayesClassifier.train(training_set)
# print("Classifier accuracy percent:", (nltk.classify.accuracy(classifier, testing_set)) * 100)
# classifier.show_most_informative_features(15)
start_time = time.time()
classifier = nltk.SklearnClassifier(RandomForestClassifier(n_estimators=4000, criterion='entropy'))
classifier.train(training_set)
print("Classifier accuracy percent:", (nltk.classify.accuracy(classifier, testing_set)) * 100)
print("%s seconds" % (time.time() - start_time))
def gender_features(word):
return {'last_letter': word[-1]}
labeled_names = ([(name, 'male') for name in names.words('male.txt')] +
[(name, 'female') for name in names.words('female.txt')])
featuresets = [(gender_features(n), gender) for (n, gender) in labeled_names]
train_set, test_set = featuresets[500:], featuresets[:500]
classifier = nltk.NaiveBayesClassifier.train(train_set)
print("Original Naive Bayes Algo accuracy percent:",
(nltk.classify.accuracy(classifier, test_set)) * 100)
classifier.show_most_informative_features(15)
MNB_classifier = nltk.SklearnClassifier(MultinomialNB())
MNB_classifier.train(train_set)
print("MNB_classifier accuracy percent:",
(nltk.classify.accuracy(MNB_classifier, test_set)) * 100)
# print(classifier.classify(gender_features('Neo')))
# print(MNB_classifier.classify(gender_features('Neo')))
BernoulliNB_classifier = nltk.SklearnClassifier(BernoulliNB())
BernoulliNB_classifier.train(train_set)
print("BernoulliNB_classifier accuracy percent:",
(nltk.classify.accuracy(BernoulliNB_classifier, test_set)) * 100)
LogisticRegression_classifier = nltk.SklearnClassifier(LogisticRegression())
LogisticRegression_classifier.train(train_set)
print("LogisticRegression_classifier accuracy percent:",
def svm(training_set,dev_test_set):
classifier = nltk.SklearnClassifier(ls()).train(training_set)
accuracy = nltk.classify.accuracy(classifier,dev_test_set)
return (classifier,accuracy)
n=1,
extended=0,
):
# make bag of words
bow = dict()
for word in nltk.ngrams(text, n):
if extended == 0:
if word not in bow:
bow[word] = True
else:
bow[word] = bow.get(word, 0) + 1
return bow
models = [
("SVM", nltk.SklearnClassifier(sklearn.svm.SVC(probability=True))),
("NaiveBayes", nltk.classify.NaiveBayesClassifier),
# ("RandomForest", nltk.SklearnClassifier(ensemble.RandomForestClassifier())),
# ("MaxEntropy", nltk.classify.MaxentClassifier),
# ("MLP", nltk.SklearnClassifier(neural_network.MLPClassifier()))
]
onlyBayes = [("NaiveBayes", nltk.classify.NaiveBayesClassifier)]
def test_classifiers(classifiers, ngrams, cutoffs):
test_counter = 0
scores = dict()
for i in range(2):
for ndx, (name, cl) in enumerate(classifiers):
for gram in ngrams:
test_results = {'tp': 0, 'fp': 0, 'tn': 0, 'fn': 0}
#main
data_list = getTrainingData()
question_list = [x for (x,y) in data_list]
label_list = [y for (x,y) in data_list]
print("\n question_list = ", question_list)
print("\n label_list = ", label_list)
vectorizer = CountVectorizer(min_df=1)
X = vectorizer.fit_transform(question_list)
print("\n X.toarray()=", X.toarray())
print("\n X.shape[0]=",X.shape[0])
test_v = vectorizer.transform(['it is a test of sales order'])
print("\n test_v.toarray()=", test_v.toarray())
training_set= [ (getContentFeature(question_list[i]), label_list[i]) for i in range(X.shape[0])]
print("\n training_set=", training_set)
myClassifier = nltk.SklearnClassifier(SVC()).train(training_set)
#Test
while True:
query = str(input("Input query:")).strip().lower()
feature = getContentFeature(query)
#print feature
print("\n result = ", myClassifier.classify(feature))