Spatio-Temporal Causal Relations at Urban Road Networks; Granger Causality Based Networks as an Insight to Urban Traffic Dynamics

Abstract

Being spatio-temporal variations of urban road traffic a critical information for understanding and predicting accurately traffic, the current paper focuses on urban road traffic dynamics understanding by introducing the notion of causality. Using 15-min aggregated travel time series from taxi GPS data, causal networks are developed. The results reveal potentials arising from mining causality beyond correlation notion among urban road paths as well as the contribution of causal networks to a decision support system for traffic management. The knowledge on causal relations and the characteristic time lags on the ‘transfer’ of the information (traffic) among the road paths is a key knowledge for traffic management, since it gives the possibility to proactively intervene in the affected road paths and to inform users for alternative routes. Being high the extendibility and transferability potentials of the proposed approach, exploitation in other transport-related problems appears promising. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.