朴素贝叶斯分类算法py

以下是一个简单的朴素贝叶斯分类器的 Python 实现。该分类器用于基于文本分类问题。

import math

class NaiveBayesClassifier:
    def __init__(self):
        self.classes = []
        self.class_word_count = {}
        self.class_total_count = {}
        self.vocabulary = set()

    def train(self, train_data):
        for class_name, text in train_data:
            if class_name not in self.classes:
                self.classes.append(class_name)
            if class_name not in self.class_word_count:
                self.class_word_count[class_name] = {}
            if class_name not in self.class_total_count:
                self.class_total_count[class_name] = 0

            for word in text.split():
                if word not in self.vocabulary:
                    self.vocabulary.add(word)
                if word not in self.class_word_count[class_name]:
                    self.class_word_count[class_name][word] = 0
                self.class_word_count[class_name][word] += 1
                self.class_total_count[class_name] += 1

    def classify(self, text):
        max_score = -math.inf
        max_class = None

        for class_name in self.classes:
            score = math.log(self.class_total_count[class_name])
            for word in text.split():
                if word in self.class_word_count[class_name]:
                    word_count = self.class_word_count[class_name][word]
                else:
                    word_count = 0
                score += math.log((word_count + 1) / (self.class_total_count[class_name] + len(self.vocabulary)))
            if score > max_score:
                max_score = score
                max_class = class_name

        return max_class

使用示例：

train_data = [
    ('spam', 'buy cheap viagra now'),
    ('spam', 'make money fast'),
    ('ham', 'hello world'),
    ('ham', 'python is awesome')
]

classifier = NaiveBayesClassifier()
classifier.train(train_data)

print(classifier.classify('buy now, make money'))
# Output: spam

print(classifier.classify('hello python world'))
# Output: ham

在该示例中，我们定义了一个简单的训练数据集，其中第一个元素表示类别，第二个元素表示文本。然后使用 `NaiveBayesClassifier` 类训练分类器，最后使用 `classify` 方法对新的文本进行分类。