Abstract:
Currently, there is an ongoing discussion regarding the role of urban planning and transport planning in the development of walkable cities. It argues for rethinking the technology-centric approach that combines urban/transport planning and technological domains, such as developing field called Geospatial Artificial Intelligence (GEOAI). This study addressed theoretical and practical challenges in walking behavior analysis. First, map pedestrian walking behavior. Second, quantifying spatiotemporal element’s impact on walking behavior is challenging. The utilization of GEOAI in this field is still deficient. The methodology of this study employs GPS-enabled location-based services to capture walking behavior and street view, isovist factors, and space syntax to quantify the environment. This method maps walking behavior using GIS and k-means clustering, an unsupervised machine-learning model used for splitting data. Additionally, Extreme Gradient Boosting (XGBoost), a supervised machine learning, is employed to analyze how spatiotemporal factors influence walking behavior. The findings highlight a significant relationship between tree view, mean depth, choice, and walking behavior. This research provides transport and urban planners with crucial insights and a novel methodological framework to develop more walkable cities, optimize urban design, transport planning strategies, and enhance urban livability and sustainability.
