2

As I know, I can just decide the number of hidden states, then put my observations fit into my HMM model, the model will generate the hidden states for me. I am really curious how it does work. Can anyone explain the calculation or share the link which has such details with me.

Pluviophile
  • 3,520
  • 11
  • 29
  • 49
Kinglun Poon
  • 21
  • 1
  • 1
    What do you mean with: "I can just decide the number of hidden states [...] the model will generate the hidden states for me?" – TitoOrt Jan 17 '20 at 13:00
  • `>>> remodel = hmm.GaussianHMM(n_components=3, covariance_type="full", n_iter=100) >>> remodel.fit(X) GaussianHMM(algorithm='viterbi', ... >>> Z2 = remodel.predict(X)` Here is the sample I found from [link](https://hmmlearn.readthedocs.io/en/latest/tutorial.html). I also worked with some sample data and they works well. As I know, number of **n_components** means number of **hidden states**. After called **fit(X)**, the **transmat_** is generated automatically, so I want to how it does work... Thanks! – Kinglun Poon Jan 20 '20 at 14:39

0 Answers0