Parameter tracking for an inventory model. (Q1412535)

This article deals with an application of hidden Markov models techniques to a discrete-time inventory model, where the demand and replinishment processes jump between different levels according to the jumps of two (`hidden') Markov chains. Using measure change techniques the problem of detecting a `best' model from a list of dynamical models tracking the behaviour of the stock level is investigated. Recursive filters which indicate the most likely inventory model are proposed.