Estimating dynamic origin-destination matrices from the ATV data

We revise the extraction of dynamic origin-destination (OD) matrices from the ATV data in public transportation systems. We extend the previous approaches by arguing that passengers plan their trips with multiple goals in mind. We argue a need for determining the multi-goal transfer trips and propose a method for recognizing such trips and including their elements in the OD matrix. We deploy the utility theory to describe the uncertainty relevant to the unobservable parameters which can impact the user's choices.