Detecting anomalous commuting patterns: Mismatch between urban land attractiveness and commuting activities

Rapid urbanization has dramatically changed the urban spatial structures, causing a mismatch between residents' commuting activities and the optimal status of the current urban facility configuration. However, limited attention has been paid to detecting these mismatched commuting patterns and their associations with built environmental characteristics. To maximize the effectiveness of urban facility allocation and improve commuting efficiency, this paper developed a framework to identify anomalous commuting interaction patterns. A weighted bipartite network considering urban land attractiveness was first constructed to analyze the commuting flows between urban units. Then a modified Hungarian algorithm was proposed to obtain the optimal commuting interaction fluxes. By comparing real and optimal interaction fluxes, two types of commuting anomalies were detected. Finally, the machine learning model was used to explore the non-linear relationships between built environment and anomalous commuting patterns. Results show the spatial distribution of areas with significant anomalous interactions and the difference between overload- and underload- related anomalous commuting patterns. Potential urban sub-centers were identified to adjust the urban spatial layouts. Besides, the nonlinear and threshold effects of the built environment on the two anomalous commuting patterns were confirmed, which can provide references for urban spatial renewal and commuting flow allocation.