possibleStates/sanitize Interaction Is Questionable #213
Description
If possibleStates is the superset of states observed in my sequence, and sanitize is TRUE, it seems that both createSequenceMatrix() and markovchainFit() both set row probabilities to 1/length(possibleStates). Is that correct? sanitize seems to force all unobserved states to communicate equally with observed states. That's not supported by my data, at least.
Rather, should it be such that, for unobserved states, the diagonal is set to 1 thereby satisfying the requirement of the new() function, i.e., that row probabilities should sum to 1? I don't think this needs code to explain but let me know if needed. Thank you!