Fit Hidden Markov Model

Fit a hidden Markov model (HMM) to sequential observation data using the Baum-Welch expectation-maximization algorithm, decode the most likely hidden state sequence via Viterbi, and select the optimal number of hidden states through information criteria.

When to Use

• You observe a sequence of emissions but the underlying generative states are not directly observable
• You suspect your data is generated by a system that switches between a finite number of regimes
• You need to segment a time series into latent phases (e.g., market regimes, speech phonemes, biological sequence annotation)
• You want to compute the probability of an observed sequence under a generative model
• You need the most likely sequence of hidden states given observations (decoding)
• You are comparing models with different numbers of hidden states for the best complexity-fit trade-off

Inputs

Required