Method for estimating origin-destination matrices using Markov models

摘要

The method of origin-destination flows estimation based on Markov properties of a transportation network is considered. A formal description of a transportation graph is provided. A canonical form of the matrix of transition probabilities for the transportation graph is obtained. Maximum likelihood estimates of transition probabilities is provided. A method for estimating indirect transition probabilities using a fundamental matrix of the Markov chain is proposed. This method is able to be used for obtaining a transport origin-destination matrix. An example of a matrix calculation of some transportation graph is presented. Opportunities for a practical usage of the proposed method are provided.