Modeling urban traffic dynamics in coexistence with urban data streams

Classic paradigm of scientific modeling is mainly based on a set of previously, accepted or assumed theories about the target phenomena and a validation procedure by limited observations. Therefore, normally data has a supporting role in the modeling process. On the other hand, recent advances in computing technology have brought us a data deluge that may change the classic paradigm of scientific modeling. Information flows and data streams have reached a level of maturity that they can play the main role in modeling of the real systems, without relying on lots of assumptions and rules in the first step. This turn may cause an inversion in the concept of modeling as a rational process. The proposed theoretical idea in this work is that traditional theory-driven models have a theoretical limit in modeling complex systems, known as curse of dimensionality and further, to highlight the fact that massive urban data streams can open up a new data-driven modeling approach, which goes beyond simple data driven analytics or eye catching info-graphics toward operational models of complex phenomena. In this work we describe a conceptual framework for modeling city wide traffic dynamics that proposes a way to encapsulate the complexity based on abstraction power of Markov chains in a coexistence with continuous data streams. Therefore, finally as an experimental set up, we applied the proposed model to a real data set, consisting of GPS traces of taxi cabs in Beijing and the results have been explained.