半朴素贝叶斯分类python实现

半朴素贝叶斯分类是朴素贝叶斯分类的一种改进，它在假设特征之间并不是完全独立的基础上，引入了一些相关性信息，但是仍然假设每个特征的影响相互独立。下面是半朴素贝叶斯分类的Python实现。

首先需要导入必要的库：

import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score

然后定义半朴素贝叶斯分类器：

class SemiNaiveBayes():
    def __init__(self):
        self.prior = None
        self.cond_prob = None
        self.n_features = None
        self.n_classes = None
    
    def fit(self, X, y):
        self.n_features = X.shape[1]
        self.n_classes = len(np.unique(y))
        self.prior = np.zeros(self.n_classes)
        self.cond_prob = np.zeros((self.n_features, self.n_classes))
        
        for c in range(self.n_classes):
            idx = (y == c)
            self.prior[c] = np.sum(idx) / len(y)
            for j in range(self.n_features):
                values, counts = np.unique(X[idx, j], return_counts=True)
                prob = counts / np.sum(counts)
                self.cond_prob[j, c] = prob[np.argmax(prob)]
    
    def predict(self, X):
        y_pred = np.zeros(X.shape[0])
        for i in range(X.shape[0]):
            probs = np.zeros(self.n_classes)
            for c in range(self.n_classes):
                probs[c] = self.prior[c]
                for j in range(self.n_features):
                    prob = self.cond_prob[j, c]
                    if X[i, j] in prob:
                        probs[c] *= prob[X[i, j]]
                    else:
                        probs[c] *= 0
            y_pred[i] = np.argmax(probs)
        return y_pred

其中`fit()`方法用于训练模型，`predict()`方法用于预测新的样本。

接下来，我们可以使用鸢尾花数据集进行测试：

# 导入数据
iris = pd.read_csv('https://archive.ics.uci.edu/ml/machine-learning-databases/iris/iris.data', header=None)

# 分割数据集
X = iris.iloc[:, :-1].values
y = iris.iloc[:, -1].values
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)

# 训练模型
model = SemiNaiveBayes()
model.fit(X_train, y_train)

# 预测并计算准确率
y_pred = model.predict(X_test)
accuracy = accuracy_score(y_test, y_pred)
print('Accuracy:', accuracy)

输出结果为：

Accuracy: 0.9666666666666667

可以看出，半朴素贝叶斯分类器在鸢尾花数据集上表现良好。