chainsaddiction.poishmm

class chainsaddiction.PoisHmm
err: bool

True, if an error was encountered during parameter estimation, else False.

m_states: int

Number of states.

n_ter: int

Number of iterations performed.

aic: float

Akaike information criterion.

bic: float

Bayesian information criterion.

llk: float

Logarithm of the likelihood.

lambda_: numpy.ndarray

Estimate of the state-dependend probabilities.

gamma_: numpy.ndarray

Estimate of the transition probability matrix.

delta_: numpy.ndarray

Estimate of the initial distribution.

lalpha: numpy.ndarray

Logarithm of the forward probabilities for each observation (rows) and state (columns).

lbeta: numpy.ndarray

Logarithm of the backwward probabilities for each observation (rows) and state (columns).

lcsp: numpy.ndarray

Logarithm of the conditional state probabilities for each observation (rows) and states (columns).

chainsaddiction.poishmm.fit(n_obs, m_states, max_iter, sdm, tpm, distr, data)
Parameters:

  • n_obs (int) – Number of observations

  • m_states (int) – Number of states

  • max_iter (int) – Maximum number of iterations

  • sdm (np.ndarray) – State-depended means

  • tpm (np.ndarray) – Transition probability matrix

  • distr (np.ndarray) – Start/initial distribution

  • data (np.ndarray) – Input data set

Returns:

Fitted HMM

Return type:

PoisHmm

Fit a HMM with Poisson-distributed states to data.

chainsaddiction.poishmm.read_params(path)
Parameters:

path (str) – Path to file

Read Poisson HMM parameters from file.