New SeoulMMOD dataset benchmarks multimodal urban mobility forecasting

Researchers have introduced SeoulMMOD, a comprehensive benchmark dataset for forecasting urban mobility in Seoul. This dataset includes three years of hourly origin-destination flow data across six transport modes, covering 25 districts and 426 sub-districts. It also incorporates supplementary data such as travel times, distances, weather, and points of interest. Initial benchmarking of 14 forecasting models revealed that while some spatio-temporal models struggle with sub-district level forecasting, naive joint multi-mode training does not consistently yield better accuracy. AI

IMPACT Establishes a reproducible testbed for multimodal, multi-year OD forecasting at city scale, potentially advancing AI applications in urban planning.

RANK_REASON The item describes a new academic dataset and benchmark for urban mobility forecasting. [lever_c_demoted from research: ic=1 ai=0.4]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• IArxiv
• ScienceCast
• Seoul
• SeoulMMOD

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →