This commit 32e5254550f4879ebe7dd6463a345a736f97098e contains a family for fitting a discrete Markov chain oscillating between 0 and 1.
However, it is probably useless, because if we have, say, p states {1, 2, ..., p} the way to fit such model (i.e. estimating the transition matrix) is via p completely separate multinomial regressions, each estimating one of the columns of the transition matrix.
But it might be interesting to implement this at some point.
mfasiolo
commented
11 months ago
This commit 32e5254550f4879ebe7dd6463a345a736f97098e contains a family for fitting a discrete Markov chain oscillating between 0 and 1.
However, it is probably useless, because if we have, say, p states {1, 2, ..., p} the way to fit such model (i.e. estimating the transition matrix) is via p completely separate multinomial regressions, each estimating one of the columns of the transition matrix.
But it might be interesting to implement this at some point.