Hello Everyone,
For example i have 3 states Markov model i.e Healthy, Sick and Death. i collected data in the last year. data form is like yes or no for each. Simple example is Healthy state having {yes,no,yes,no,yes,no,no,yes} , Sick={ no,yes,no,es..} and Death={no,yes,no,yes,yes..}. and after one again i collected data in the form of these three states.So my doubt is how to find state transition probabilities when data format like above example.
Thank-you
Suvarna Narendra : I won't state obvious things here or funny remarks. I think you have a valid question and I'll try to answer it.
Your data is perhaps unsuitable for Markov Chain modelling. Either you work with probabilities to change states, e.g., a chance of recovery from Sick to Health where you somehow computed or have the number from data or the time it took for Sick patients to return to Health state, then with the time you convert to a rate over a period of time.
In the first case, you are working with a DTMC, Discrete Time Markov Chain whereas the second case it is a CTMC, Continuous Time Markov Chain. They both have methods and tools to aid you come up with state permanence probabilities and add to your analysis.
Your data reflects the final state of patients. You should work with other data to have parameters to your model.
I urge you to read my book on Markov Chains (it is devoted to programming, but I think you'll appreciate the underlying comments and discussions on modelling).
Link: https://figshare.com/articles/book/Markov_Chains_for_programmers/19714384/2
And on ResearchGate: Book Markov Chains for programmers
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix