Who can give me some practical advice/examples on how to implement a Hidden Markov Model. I would like to use that in the traffic engineering context under various traffic flow conditions. E.g. on the approach to a queue, does driver maintain or slow down when he sights a queue or red light? From the literature I have one thing that seems to elude me is how to calculate or estimate the transition probability matrix.
Hello. This is a well treated subject in many fields. HMMs were introduced into electrical engineering in the 1980s - Rabiner's 1989 Proc. IEEE paper is a good place to start. I previously wrote a short survey of HMM approaches for radar tracking (please refer to the link below). It's a bit old (!) but covers the basics.
As far as transition probabilities go, you have to have some idea what they are (or use a uniform distribution). In some problems you just specify them in others you wish to learn them using the Baum-Welch re-estimation procedure, which might require quite a lot of data. An example of this method is in the 2nd link below (dealing with a cell biology application).
For your appication, it seems like you wish to model driver behaviour (speed?) based on observed traffic conditions. You must first say what states your HMM has then what the probability of stating the the same state or moving to another is. This is where the domain knowledge is important and I cannot help you with the traffic engineering field apart from recommending a google search.
Article A Survey of Hidden Markov Model Tracking with Emphasis on OTHR
Article Evaluation and Estimation of Various Markov Models with Appl...
You may look on these papers
http://www.nrso.ntua.gr/geyannis/images/stories/ge/Publications/Papers-Conferences/geyannis-pc71.pdf
http://link.springer.com/chapter/10.1007/978-3-642-33478-8_48#page-1
http://www.sciencedirect.com/science/article/pii/S0968090X14000394
http://etd.lsu.edu/docs/available/etd-04232010-092757/unrestricted/Dissertation.pdf
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size