mhmmparam-class

<dl class="dl-horizontal">
 <dt>K</dt><dd>numeric. Number of classes.</dd>
 <dt>M</dt><dd>numeric. Number of latent states (activity levels).</dd>
 <dt>A</dt><dd>list. Matrices of the transition probailities for each class.</dd>
 <dt>delta</dt><dd>numeric. Proportions of the classes.</dd>
 <dt>pi</dt><dd>matrix. Probabilities of the latent states per class (stationary distribution of the Markov chains).</dd>
 <dt>lambda</dt><dd>list. Parameters of the emission laws (eps: proportions of the zero-inflated component, a: shapes of the gamma distributions, b: rates of the gamma distributions)</dd>
</dl>

Estimation of the latent states and partition by maximum likelihood. Model can be used for analyzing accelerometer data. In such a case, the latent states corresponds to activity levels and the partition permits to consider heterogeneity within the population. Emission laws are zero-inflated gamma distributions. Their parameters depends on the latent states but not on the partition, to compare the time spent by activity levels between classes. Model description is available in Du Roy de Chaumaray, M. and Marbac, M. and Navarro, F. (2019) <arXiv:1906.01547>.

Matthieu Marbac

MHMM

Finite Mixture of Hidden Markov Model

mhmmparam-class function

<dl class='dl-horizontal'>
 <dt>K</dt><dd>numeric. Number of classes.</dd>
 <dt>M</dt><dd>numeric. Number of latent states (activity levels).</dd>
 <dt>A</dt><dd>list. Matrices of the transition probailities for each class.</dd>
 <dt>delta</dt><dd>numeric. Proportions of the classes.</dd>
 <dt>pi</dt><dd>matrix. Probabilities of the latent states per class (stationary distribution of the Markov chains).</dd>
 <dt>lambda</dt><dd>list. Parameters of the emission laws (eps: proportions of the zero-inflated component, a: shapes of the gamma distributions, b: rates of the gamma distributions)</dd>
</dl>

mhmmparam-class: Constructor of `'>mhmmparam` class

Description

Arguments

Examples