Fit MHMM with Clusters of different rows about seqhmm HOT 1 CLOSED

Hi xlko,

Yes, clusters can be of varying sizes (different numbers of subjects). I'm assuming there is some confusion with the terminology here. The example you are referring to is for multichannel sequence data meaning that for each subject we have multiple parallel sequences; in the example we have three-channel data with information on marriage, parenthood, and residence annually for each individual.

I'm assuming you have "single-channel" data, i.e., one sequence per subject. The data you give to the observations argument should be an stslist object of the form N x T where N is the number of subjects and T the the length of the sequence (typically the number of time points). If you have multichannel data you have a list of stslists where each list element corresponds to one channel.

I urge you to read the seqHMM vignette which explains the terminology in more detail and shows some examples. See also the help file of the build_hmm function which starts with an example to estimating an HMM to single-channel data.

from seqhmm.