本文整理汇总了Python中DataModel.generate_word_probabilites方法的典型用法代码示例。如果您正苦于以下问题：Python DataModel.generate_word_probabilites方法的具体用法？Python DataModel.generate_word_probabilites怎么用？Python DataModel.generate_word_probabilites使用的例子？那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在类DataModel的用法示例。
在下文中一共展示了DataModel.generate_word_probabilites方法的1个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于系统推荐出更棒的Python代码示例。
示例1: NaiveBayesClassifier
# 需要导入模块: import DataModel [as 别名]
# 或者: from DataModel import generate_word_probabilites [as 别名]
class NaiveBayesClassifier(object):
def __init__(self, training_file, testing_file):
super(NaiveBayesClassifier, self).__init__()
self.training_file = training_file
self.testing_file = testing_file
self.model = None
self.time_to_train_string = ""
self.time_to_label_string = ""
self.classification_accuracy_string = ""
self.training_accuracy_string = ""
def train(self):
start = time.clock()
reviews = open(self.training_file, "r")
word_count = [[], []]
# Populate word_count array with 2 arrays, word_count[0] is for the 0 reviews and word_count[1] is
# for the 1 reviews. Each element of those arrays contain a dictionary for the word count for a specific review
ReviewParser.parse_for_training(reviews, word_count)
self.create_model(word_count)
end = time.clock()
self.time_to_train_string = str(int(end - start)) + " seconds (training)"
def classify(self):
start = time.clock()
reviews = open(self.testing_file, "r")
parsed_reviews = ReviewParser.parse_for_labeling(reviews)
classified_reviews = self.classify_reviews(parsed_reviews)
# This method will print the labels followed by a newline
self.get_classification_accuracy(classified_reviews)
self.run_classification_on_training()
end = time.clock()
self.time_to_label_string = str(int(end - start)) + " seconds (labeling)"
# Print all the required timing and accuracy
print(self.time_to_train_string)
print(self.time_to_label_string)
print(self.training_accuracy_string)
print(self.classification_accuracy_string)
# Helper Methods
# Training
def create_model(self, review_word_counts):
self.model = DataModel(len(review_word_counts[1]), len(review_word_counts[0]))
self.model.generate_word_probabilites(review_word_counts[1], review_word_counts[0])
# Classification
def classify_reviews(self, parsed_reviews):
classified_reviews = []
for review_info in parsed_reviews:
prob_neg = math.log10(self.model.probability_is_negative_review)
prob_pos = math.log10(self.model.probability_is_positive_review)
# Use the sum of logs to deal with really small numbers log(P(C=c)) + sum_i(P(X=x_i|C=c))
for word in review_info["words"][1:]:
if word in self.model.word_probabilites["negative"]:
conditional_neg = self.model.word_probabilites["negative"][word]
if conditional_neg > 0:
prob_neg += math.log10(conditional_neg)
if word in self.model.word_probabilites["positive"]:
conditional_pos = self.model.word_probabilites["positive"][word]
if conditional_pos > 0:
prob_pos += math.log10(conditional_pos)
classified_reviews.append(
{
"real_category" : int(review_info["real_category"]),
"calculated_category" : 1 if prob_pos > prob_neg else 0
}
)
return classified_reviews
def get_classification_accuracy(self, classified_reviews):
num_correct = 0
total = 0
for review_classification in classified_reviews:
total += 1
print(review_classification["calculated_category"])
if review_classification["real_category"] == review_classification["calculated_category"]:
num_correct += 1
classification_accuracy = float(num_correct) / float(total)
self.classification_accuracy_string = str(classification_accuracy) + " (testing)"
def get_training_accuracy(self, classified_reviews):
num_correct = 0
total = 0
for review_classification in classified_reviews:
total += 1
if review_classification["real_category"] == review_classification["calculated_category"]:
num_correct += 1
training_accuracy = float(num_correct) / float(total)
self.training_accuracy_string = str(training_accuracy) + " (training)"
def run_classification_on_training(self):
reviews = open(self.training_file, "r")
parsed_reviews = ReviewParser.parse_for_labeling(reviews)
classified_reviews = self.classify_reviews(parsed_reviews)
self.get_training_accuracy(classified_reviews)