Benchmarking living-zone plans with mobility community detection: Evidence from Seoul's mobile-phone-based mobility data

초록

Planning self-contained neighbourhoods has emerged as a strategy to enhance walkability and mitigate traffic congestion in cities worldwide. Yet official living-zone boundaries are often drawn through expert judgement and administrative constraints, leaving their empirical alignment with everyday mobility largely untested. This study proposes a mobility-network-based evaluation framework that benchmarks official living-zone plans against data-derived mobility communities. Because living-zone planning concerns routine access to shopping, leisure, social, cultural, and public services rather than labour-market catchments, we use anonymised daytime non-work/non-school origin-destination flows for Seoul (424 dongs) from the Seoul Living Mobility Dataset v3.3 (SLMD). We derive mobility communities with two widely used community-detection algorithms (Louvain and Leiden) and evaluate the resulting zones with modularity, internal flow ratio (IFR), spatial contiguity, and intersection-over-union (IoU) relative to Seoul's official 116 living zones. Results show that algorithm choice matters for planning usability: in our case study, Leiden more reliably matches the policy target of 116 zones and yields fewer spatially fragmented clusters than Louvain under comparable parameter tuning. While many official zones closely align with mobility communities, several districts display systematic mismatches with lower cohesion and self-containment. Planning implications are derived by jointly interpreting IoU, modularity gaps, and IFRk gaps as relative priority signals rather than universal thresholds, which can help prioritise boundary review, local connectivity improvements, and service-area adjustments. By shifting the focus from merely 'drawing a better map' to auditing and iteratively updating living-zone plans, the study demonstrates how mobility data can support evidence-based, adaptive neighbourhood planning.

키워드