fork download 
  1. import numpy as np
  2.  
  3. class HiddenMarkovModel:
  4.     def __init__(self, states, observations, start_prob, trans_prob, emit_prob):
  5.         self.states = states
  6.         self.observations = observations
  7.         self.start_prob = start_prob
  8.         self.trans_prob = trans_prob
  9.         self.emit_prob = emit_prob
  10.  
  11.     def viterbi(self, obs_seq):
  12.         n_states = len(self.states)
  13.         T = len(obs_seq)
  14.  
  15.         dp = np.zeros((n_states, T))
  16.         path = np.zeros((n_states, T), dtype=int)
  17.  
  18.         for s in range(n_states):
  19.             dp[s, 0] = self.start_prob[s] * self.emit_prob[s][self.observations.index(obs_seq[0])]
  20.             path[s, 0] = 0
  21.  
  22.         for t in range(1, T):
  23.             for s in range(n_states):
  24.                 prob_transition = dp[:, t-1] * self.trans_prob[:, s]
  25.                 best_prev_state = prob_transition.argmax()
  26.                 dp[s, t] = prob_transition[best_prev_state] * self.emit_prob[s][self.observations.index(obs_seq[t])]
  27.                 path[s, t] = best_prev_state
  28.  
  29.         best_last_state = dp[:, T-1].argmax()
  30.         best_path = [best_last_state]
  31.  
  32.         for t in range(T-1, 0, -1):
  33.             best_last_state = path[best_last_state, t]
  34.             best_path.insert(0, best_last_state)
  35.  
  36.         best_state_sequence = [self.states[state] for state in best_path]
  37.         return best_state_sequence
  38.  
  39.  
  40. if __name__ == "__main__":
  41.     states = ['S1', 'S2']
  42.     observations = ['obs1', 'obs2', 'obs3']
  43.  
  44.     start_prob = np.array([0.6, 0.4])
  45.     trans_prob = np.array([[0.7, 0.3],
  46.                            [0.4, 0.6]])
  47.     emit_prob = np.array([[0.5, 0.4, 0.1],
  48.                           [0.1, 0.3, 0.6]])
  49.  
  50.     obs_seq = ['obs3', 'obs1', 'obs3']
  51.  
  52.     hmm = HiddenMarkovModel(states, observations, start_prob, trans_prob, emit_prob)
  53.     best_sequence = hmm.viterbi(obs_seq)
  54.  
  55.     print("Most likely state sequence:", best_sequence)
  56.  
  57. # your code goes here
Success #stdin #stdout 0.79s 41548KB
comments (?)
stdin 
Standard input is empty
stdout 
Most likely state sequence: ['S2', 'S1', 'S2']
https://ideone.com/42HyWB
language:
Python 3 (python  3.12)
created:
6 days ago
visibility:
public

Share or Embed source code

Discover > Sphere Engine API

The brand new service which powers Ideone!

Discover > IDE Widget

Widget for compiling and running the source code in a web browser!