维特比算法

import numpy as np
invisible = ['Sunny','Rainy']
activity = ['walk', 'shop', 'clean']
pi = [0.4, 0.6]
trainsion_prob = [[0.3, 0.7],[0.6, 0.4]]
emission_prob = [[0.6, 0.3, 0.1], [0.1, 0.4, 0.5]]
obs_seq=[0,1,2]
# 最后返回一个Row*Col的矩阵结果

 
def viterbi(transiton_prob,emit_prob,pi,obs_seq):
    # 转换为矩阵
    tansition_prob=np.array(transiton_prob)
    emit_prob=np.array(emit_prob)
    pi=np.array(pi)
    obs_seq=[0,1,2]

    Row=tansition_prob.shape[0] #row等于转移矩阵的行数
    Col=len(obs_seq)
    F=np.zeros((Row,Col))
   
    #emit_prob[:,obs_seq[0]]表示从初始状态转移到观测到的第一个状态的转移概率
    F[:,0]=pi*np.transpose(emit_prob[:,obs_seq[0]])
    # 遍历观测序列中的每一个值
    for t in range(1,Col):
        list_max=[]
        for n in range(Row):
            list_x=list(np.array(F[:,t-1])*np.transpose(tansition_prob[:,n]))    
            #获取最大概率
            list_p=[]
            for i in list_x:
                list_p.append(i)
            list_max.append(max(list_p))
        F[:,t]=np.array(list_max)*np.transpose(emit_prob[:,obs_seq[t]])
    return F

F = viterbi(trainsion_prob, emission_prob, pi, obs_seq)
print(F)
i = 0
for i in range(3):
    if (F[0][i] > F[1][i]):
        print("Sunny")
    else:
        print("Rainy")

[[0.24     0.0216   0.004032]
 [0.06     0.0672   0.01344 ]]
Sunny
Rainy
Rainy