ABSTRACT Transit flows can provide useful insights into urban mobility patterns and dynamics. Generally, the transit flows between the origin and destination locations are aggregated into areal units for statistical analysis. A well‐known issue is the modifiable areal unit problem (MAUP), where its sensitivity depends on the basic areal units. This study addresses the MAUP of aggregated flows by areal units, which are designed along the public transit network. A public transit flow model based on the quasi‐Poisson regression is introduced and systematically assessed in terms of different scales and zones for transit flow aggregation. The proposed model for public transit flows incorporates network topological measures, such as degree, betweenness, and PageRank centralities, which are significant explanatory factors for trips. Results indicate that the MAUP significantly impacted various statistical outputs, including regression coefficients and p values for public transit flow modeling. Moreover, the sensitivity of the MAUP varies depending on how the transit flows are aggregated. Our findings contribute to a deeper understanding of flow data aggregation issues based on the network structure. The significance of network topological measures is revealed in public transit flow modeling. The experimental results can help improve public transit flow prediction and establish effective public transportation policies.
Kim et al. (Mon,) studied this question.