A Revised Snake Clustering Method for Static Partitioning of Urban Road Networks

Partitioning of urban road networks is crucial for alleviating traffic congestion in large cities. In this paper, a revised clustering algorithm for road network partitioning is developed based on an existing snake clustering method. In response to the issues present in the existing snake clustering method, the revised snake clustering (RSC) method proposed in this paper tackles these challenges through targeted improvements. To address the significant randomness in the initialization, RSC method selects appropriate key road segments as the deterministic starting state. During computational process, RSC method combines the direction constraint angle with an adaptive threshold to control the expansion direction and termination condition of the snake sequence. Additionally, to overcome the simplicity of the traditional similarity measurement, RSC method comprehensively integrates temporal and spatial factors to constructs a spatiotemporal matrix as a new similarity measurement. From the simulation results of applying the proposed algorithm to a speed dataset collected from a road network in the northeast of Beijing Second Ring Road, the revised snake clustering method outperforms conventional methods, achieving 17.4% higher cluster homogeneity and 17.9% enhanced inter-cluster heterogeneity in each sub-region.