Characterizing Traffic Speed Distribution around Scenic Spots with Gaussian Mixture Model

This study proposes an approach based on the Gauss Mixture Model with the Expectation Maximization algorithm to effectively characterize traffic speed distributions around scenic spots. First, a short description of the data source and associated road conditions around the scenic spots in Yangzhou, China, are given. Next, the fluctuation characteristics of traffic speed are analyzed and compared under three patterns, i.e., weekdays, weekends, and national holidays. In contrast, the traffic speed distributions are fitted by the Gauss Mixture Model (GMM) based on the Expectation Maximization algorithm. Finally, the detailed characteristics of traffic speed distributions are investigated. Results show that the proposed approach can characterize traffic speed distributions around scenic spots. The findings can help traffic authorities make proper congestion mitigation strategies and assist drivers in making reliable vacation travel plans to avoid traffic congestion around scenic spots.