Modeling hiking speed via interpretable machine learning: Uncovering nonlinear impacts of multidimensional environmental factors
• Slope is the most influential variable affecting hiking speed in Hong Kong trails. • LightGBM best predicts segment speed from multi-source geodata. • Road type, altitude, and precipitable water show significant nonlinear impacts on mobility. • SHAP analysis uncovers the threshold and interaction among environmental variables. • Findings provide actionable insights for trail management and weather-responsive planning.
Authors
- Botao Zhang (ORCID: https://orcid.org/0000-0003-3691-7876)
- Jin Yeu; id_orcid 0000-0002-7682-4327 TSOU
- Rui Ma (ORCID: https://orcid.org/0000-0003-4153-9101)
- Hao Zheng (ORCID: https://orcid.org/0009-0007-1482-1116)
- Haochen Shi (ORCID: https://orcid.org/0000-0001-8634-3992)
- Hao Huang (ORCID: https://orcid.org/0000-0001-9801-5929)
Institutions
- Hong Kong Polytechnic University (HK)
- City University of Hong Kong (HK)
- Guangzhou University (CN)
- Zhengzhou University (CN)
Publication Details
- Journal
- Applied Geography
- Published
- 2026-05-27
- DOI
- https://doi.org/10.1016/j.apgeog.2026.104052
- Primary Topic
- Hydrological Forecasting Using AI
- Type
- article
- Field-Weighted Citation Impact
- 0.00