How do we interpret hidden States in a latent Markov model?


I have used LMM model multiple times and do understand the calculations beneath. But find it really hard to explain hidden States to business owners. Can someone give me a simpler explanation of these probabilistic events?


Hidden state in a LMM can be approximated as Principal components of the ownership flag you have on each time point. For instance, if you have 10 ownerships and 4 hidden state, this scenario is a analogous to 10 variable 4 Principal components. You now will be able to model a sequence of states which does not satisfy Markov property by inducing this latent state. The same way you can use a logistic regression even with collinear variable by inducing Principal components.

Hope this helps.


I came across Markov Models and found it is very useful concept to apply in Banking and Insurance Domain. I was just wondering, where all we can apply in Insurance domain! Please anyone can share, If you have any example or link on this model to apply in insurance domain.

Thank you in advance.