Exploring the recreational micromobility in relation to historic urban areas and social media: Insights from machine learning approaches

Abstract

Recreational micromobility exhibits characteristics of sustainable travel behavior and recreational activities, offering fresh perspectives for landscape and urban studies. The 5D framework of the built environment − Density, Diversity, Design, Destination Accessibility, and Distance to Transit − are thus often used to examine their influence on recreational micromobility. However, the relationships between recreational micromobility, historic urban areas as a distinct type of built environment, and the shaping role of social media remain underexplored. To address this research gap, we first integrate 5D, History and Cultural Heritage (H), and Social Media (S) dimensions to develop an innovative HS-5D framework to guide the exploration of recreational micromobility. Then, we conduct a Guangzhou (China) case study using seven machine learning models with large-volume GPS trajectories and multisource data. Our findings highlight that (1) the HS-5D framework has 10% higher explanatory power than the traditional 5D framework when examined by the Light Gradient Boosting Machine model with the Bayesian optimization algorithm; (2) among all dimensions, the Destination Accessibility has the most significant impact on recreational micromobility, followed by the Social Media and the Historical and Cultural Heritage; (3) the Density and Diversity dimensions have a negative impact on recreational micromobility; (4) when a place has a low rating in social media, high accessibility tends to decrease recreation micromobility of individuals. Our study provides insights into understanding the complex relationships between recreational micromobility, historic urban areas and social media better and helps develop strategies that improve the quality of urban life in a sustainable manner.