Modelo de predição de escolha de rotas baseado em traces de mobilidade veicular

Abstract

The study of vehicular mobility and its impact on the development of cities is an important research topic, contributing to the scenario of Intelligent Transport Systems (ITS). In this work, we introduce a multinomial logit model (MNL) to predict which route a user will take given a set of possible routes between origin and destination points and other contextual information. For this, we define and apply a statistical approach to understand how vehicles behave in an urban setting considering the effects of traffic. The model is applied to two real vehicle mobility tracks and the results show that it is able to capture the influences of existing factors, obtaining better predictions compared to two baselines. The results indicate that the model can be applied to fill in large spatial gaps, such as the generation of synthetic but realistic trajectory data for origin-destination tracks.