朴素贝叶斯

输入和输出

极大似然估计法

算法流程

import numpy as np

def createData( ):
    #训练数据集
    T=np.array([ [[1,'S'],-1], [[1,'M'],-1], [[1,'M'],1], [[1,'S'],1],
                 [[1,'S'],-1], [[2,'S'],-1], [[2,'M'],-1], [[2,'M'],1],
                 [[2,'L'],1], [[2,'L'],1], [[3,'L'],1], [[3,'M'],1],
                 [[3,'M'],1], [[3,'L'],1], [[3,'L'],-1] ])
    #X1,X2为特征，取值的集合分别为A1,A2
    #Y为类标记，Y包含于C
    X1 = [ 0 ] * len(T)
    X2 = [ 0 ] * len(T)
    C = [ 0 ] * len( T )
    for i in range(len(T)):
        X1[i]=T[i][0][0]
        X2[i]=T[i][0][1]
        C[i]=T[i][1]
        
    A1=list(set(X1))
    A2=list(set(X2))
    Y=list(set(C))
    return X1, X2, C, A1, A2, Y
    
def probability( X1, X2, C, A1, A2, Y ):
    P_y={}  #P(y)先验概率
    for y in Y:
        P_y[y] = C.count(y)/len(C)
    
    P_xy={}     #P(xy)
    X=A1.copy()
    X.extend(A2)
    for y in Y:
        for x_index in range(len(X)):
            count=0
            x=X[x_index]
            key = str(x) + ',' + str(y)
            if x in A1:
                for i in range(len(X1)):
                    if X1[i] == x and C[i] == y:
                        count=count+1
            elif x in A2:
                for j in range(len(X2)):
                    if X2[j] == x and C[j] == y:
                        count=count+1
            P_xy[key]=count/len(C)

    P={}    #条件概率P(x|y)
    for x in X:
        for y in P_y.keys():
            key = str(x) + '|' + str(y)
            P[key] = P_xy[str(x)+',' +str(y)] / P_y[y]
    
    return P_y,P

def testProbability( x ):
    X1, X2, C, A1, A2, Y=createData()
    P_y,P=probability( X1, X2, C, A1, A2, Y )    
    testP=[ 0 ] *(len(Y))
    i=0
    for y in Y:
        key1 = str(x[0]) + '|' + str(y)
        key2 = str(x[1]) + '|' + str(y)
        testP[i]=P_y[y]*P[key1] * P[key2]
        i=i+1

    print(Y[testP.index(max(testP))])

testProbability( [2,'S']  ) #实例[2,S]分类

**

贝叶斯估计法

**


其中Sj是X的可能取值数，K是Y的可能取值数

import numpy as np

def createData( ):
    n=1  #拉普拉斯平滑
    #训练数据集
    略
    return n,X1, X2, C, A1, A2, Y
    
def probability( n,X1, X2, C, A1, A2, Y ):
    P_y={}  #P(y)先验概率
    for y in Y:
        P_y[y] = (C.count(y)+n)/(len(C)+(len(Y)+1)*n)
    
    N_xy={}     #P(xy)
   略

    P={}    #条件概率P(x|y)
    for x in X:
        for y in P_y.keys():
            key = str(x) + '|' + str(y)
            if x in X1:
                de= C.count(y) + len(A1) + 1
            elif x in X2:
                de= C.count(y) + len(A2) + 1
            P[key] = (N_xy[str(x)+',' +str(y)]+1) / de
    
    return P_y,P

def testProbability( x ):
   略

testProbability( [2,'S']  ) #实例[2,S]分类