On Processing of Correspondence Matrices in Transport Systems

The paper considers the analysis of data in so-called origin-destination matrices. Also in relation to them often use the term matrix of correspondence. As practical examples, we are using data from Moscow metro and Moscow suburb railways. The correspondence matrix shows the number of trips between two points (stations) for a selected time interval. Theoretically, such data being calculated for the given time intervals describe all the characteristics of the passenger load. But practically, it depends on the existing models for traffic. Currently, such data used, mostly, just to calculate simple statistics. But such a simplified approach completely ignores the temporal characteristics of the traffic. In this paper, we discuss data mining models and approaches for origin-destination matrices analysis. Also, we describe the results of applying such approaches to analyzing data from the Moscow region.