Spatial-temporal correlations among the data play an important role in traffic flow prediction. Correspondingly, traffic modeling and prediction based on big data analytics emerges due to the city-scale interactions among traffic flows. A new methodology based on sparse representation is proposed to reveal the spatial-temporal dependencies among traffic flows so as to simplify the correlations among traffic data for the prediction task at a given sensor. Three important findings are observed in the experiments: (1) Only traffic flows immediately prior to the present time affect the formation of current traffic flows, which implies the possibility to reduce the traditional high-order predictors into an 1-order model. (2) The spatial context relevant to a given prediction task is more complex than what is assumed to exist locally and can spread out to the whole city. (3) The spatial context varies with the target sensor undergoing prediction and enlarges with the increment of time lag for prediction. Because the scope of human mobility is subject to travel time, identifying the varying spatial context against time lag is crucial for prediction. Since sparse representation can capture the varying spatial context to adapt to the prediction task, it outperforms the traditional methods the inputs of which are confined as the data from a fixed number of nearby sensors. As the spatial-temporal context for any prediction task is fully detected from the traffic data in an automated manner, where no additional information regarding network topology is needed, it has good scalability to be applicable to large-scale networks.

A new methodology based on sparse representation is proposed to detect the relevant sensors for traffic flow prediction at a given sensor. It performs remarkably better than the least square fitting and the local spatial context based methods. Some interesting phenomena have been observed in the experiments: (1) In general, hundreds of sensors distributed on the whole road network are relevant to a prediction task, which implies a much wider correlation range than what was assumed previously. (2) The number of relevant sensors is subject to the targeted sensor undergoing prediction due to location-specific network topology. (3) The spatial correlation scale increases with the increment of time lag while the performance degradation is less than that of the local spatial context based methods. As the scope of human mobility is subject to time lag, identifying the varying spatial context against time lag is crucial for prediction.