(李航统计方法学习）朴素贝叶斯Python实现

朴素贝叶斯属于生成模型，学习数据概率分布P(X,Y)，然后求后验概率P(Y|X)。对条件概率分布作条件独立性假设。
模型：贝叶斯定理
策略：后验概率最大化（等价于期望风险最小化）
算法：略
朴素贝叶斯在进行概率估计时有两种方式：基于最大似然估计、基于贝叶斯估计。朴素贝叶斯可以进一步扩展成贝叶斯网络

import numpy as np

def Train(X_train,Y_train,feature):
    global  class_num,label
    class_num=2
    label=[1,-1]
    feature_len=3
    feature=[[1,'S'],[2,'M'],[3,'L']]
    prior_probability=np.zeros(class_num)
    conditional_probability=np.zeros((class_num,feature_len,2))
    pos,neg=0,0
    for i in range(len(Y_train)):
        if Y_train[i] == 1:
            pos+= 1
        else:
            neg += 1
    ##计算出P(Y)
    prior_probability[0]=pos/len(X_train)
    prior_probability[1]=neg/len(X_train)
    ##统计P(X1,X2|Y)，假设X1与X2相互独立，计算P(X1,X2|Y)=P(X1|Y)*P(X2|Y),所以统计X1,X2不同取值对应的不同Y的数量
    for i in range(class_num):
        for j in range(feature_len):
            for k in range(len(Y_train)):
                if Y_train[k]==label[i]:
                    if X_train[k][0]==feature[j][0]:
                        conditional_probability[i][j][0]+=1
                    if X_train[k][1]==feature[j][1]:
                        conditional_probability[i][j][1]+=1
    class_label_num=[pos,neg]
##计算P(X1,X2|Y)
    for i in range(class_num):
        for j in range(feature_len):
            conditional_probability[i][j][0]/=class_label_num[i]
            conditional_probability[i][j][1]/=class_label_num[i]
    return  prior_probability,conditional_probability

def Predict(X_test,prior_probability,conditional_probability,feature):
    result=np.zeros(len(label))
    for i in range(class_num):
        fea0,fea1=0,0
        for j in range(len(feature)):
            if feature[j][0]==X_test[0]:
                fea0=conditional_probability[i][j][0]
            if feature[j][1]==X_test[1]:
                fea1=conditional_probability[i][j][1]
        result[i]=fea0*fea1*prior_probability[i]
    result=np.vstack([result,label])
    return result
def main():
    X_train=[[1, 'S'], [1, 'M'], [1, 'M'], [1, 'S'],  [1, 'S'],
               [2, 'S'], [2, 'M'], [2, 'M'], [2, 'L'],  [2, 'L'],
               [3, 'L'], [3, 'M'], [3, 'M'], [3, 'L'],  [3, 'L']]
    Y_train = [-1, -1, 1, 1, -1, -1, -1, 1, 1, 1, 1, 1, 1, 1, -1]
    feature = [[1, 'S'],
               [2, 'M'],
               [3, 'L']]

    testset = [2, 'S']

    prior_probability, conditional_probability = Train(X_train, Y_train, feature)

    result = Predict(testset, prior_probability, conditional_probability, feature)
    print(result)


if __name__ == '__main__':
    main()