a method of urban traffic flow speed estimation using sparse floating car data

The sample spatio-temporalsparsity is one of the major challenges for traffic estimation when using floating car data (FCD).Spatio-temporal characteristics of road traffic flow are analysed and applied to build a naive Bayes-based traffic estimation model which is proposed to estimate the missing traffic state of the roads which are not covered by samples. In the model, the adjacent period traffic flow speed of the target road segment is considered for the representation of the time characteristic. And instead of Euclidean distance and topology relationship, urban traffic flow similarity relationships are mined to quantify the interior space features of urban traffic.The result demonstrates that the method is effective for missing traffic state estimation and more precision compared to traditional methods based on topology relationship.As a conclusion, the proposed model can solve the spatio-temporal sparsity problem effectively, which has a strong practical significance for traffic application based on FCD.