Improving Road Traffic Prediction By using Dependencies Study: Cross-Correlation based Approach

Traffic control and management on a urban road network become more complex face to exponential growth in the volume of cars and road segments. To meet the growing demand of accurate traffic prediction, the study of relationship between road segments is necessary before prediction calculation. Correlation theory has been well developed to provide a better interpretation of dependency for understanding how Time series are related in multivariate model(MV). Pearson Coefficients based Cross-Correlation calculation is proposed to detect dependency between traffic segments in large scale road network modeled by MVTS, at this stage, dependency extraction allows to limit the use of only data collected from points related to a target point to be predicted, For our study we use a 6th road ring as most crowded area of Beijing. An interpretation of results are provided as Scatter-plot.