Optimal Selection of Clustering Analysis Model for Traffic Congestion Patterns

Based on the four parts of clustering analysis, three clustering analysis models of traffic congestion patterns were established. In order to verify the accuracy of the three clustering analysis models of traffic congestion pattern and select the best model, evaluation indexes should be found to conduct quantitative evaluation on the three models. First, this paper proposes three evaluation indexes according to the results of cluster analysis, “coefficient of intra-class variation”, “coefficient of inter-class variation”, and “coverage”. Then, three models were used to evaluate the clustering results of the data in the second quarter of 2018, and a relatively superior model was obtained. Finally, the model is applied to the data of the same period in 2019 to verify its stability. The results show that model III is the best model for clustering analysis of traffic congestion patterns. By comparing the data of the same period in 2019, it is shown that the model is reliable and stable.