Estimating Bus Speed Distribution of Bimodal Traffic Using Vine Copula

Accurate estimation of bus speed can improve urban mobility by helping passengers plan their trips better. However, it is difficult to estimate bus speed because of the interaction of social vehicles and buses in the road network. In this paper, a vine copula-based approach is proposed to model conditional probability distribution of bus speed by accounting for cars correlation. The marginal distributions of car speed and bus speed of consecutive segments along on arterial road are estimated by fusing multi-resource data. The D-vine copula model is introduced to model the dependent structure of bimodal traffic speed in the adjacent segments between bus stops. Moreover, the conditional probability distribution curves are estimated based on the D-vine copula model. The simulated results illustrate that the proposed D-vine copula model is applicable for revealing complex correlation between buses and cars using the corresponding conditional probability distribution of bus speed.