Urban vehicular traffic: fitting the data using a hybrid stochastic model. Part II

TL;DR

This paper analyzes urban vehicular traffic using a hybrid stochastic model, validating its effectiveness across multiple days and proposing rule changes to optimize flow and reduce idle times.

Contribution

It introduces a hybrid stochastic model combining linear and nonlinear components to accurately describe urban traffic dynamics and tests its applicability over different days.

Findings

The hybrid model fits traffic data well across multiple days.

Parameters fixed for one day can predict other days' traffic.

Proposed rule changes can improve traffic flow and reduce idle times.

Abstract

In this second part of our research we used the models presented in \emph{Modeling a vehicular traffic network. Part I} \cite{ogm1} to perform an analysis of the urban traffic as recorded by cameras distributed in a chosen sector of Tigre, a city in the province of Buenos Aires, Argentina. We found that the circulation of vehicles -- the traffic dynamics --, along a whole day, can be described by a hybrid model that is an adapted blend of model 2, for an open linear system, with model 3, which is nonlinear, developed in Part I. The objectives of this work were, firstly, to verify whether the vehicular flux can be modeled as an $n$-step stochastic process for its evolution, $n$ for the time. Secondly, to find out if the model, with its parameters fixed to describe the traffic of a single day, may adequately describe the traffic in other days. Thirdly, to propose changes in the already established set of the urban traffic rules in order to optimize the vehicular flow and to diminish the average time that a vehicle stays idle at the semaphores. We estimate that the goals were achieved satisfactorily within the margins of the experimental errors of the gathered data.