Abstract
Leisure walking (including running) is the easiest way to combat lack of physical activity of urban residents, which is declared as the third leading cause of death by World Health Organisation. Hence, understanding what drives people for a leisure walk is vital for the design and planning of healthy and vibrant neighbourhoods. In this work we investigate and identify which (urban) features can explain the spatial distribution of leisure walk amounts, using a collection of more than 30,000 leisure walks collected from fitness tracking apps in Singapore. First, we conduct a spatial analysis of leisure walk data using a grid-based network of Singapore and examine features, including land-use mix, street typology, greenness and the presence of particular facilities, such as bus stops and traffic lights. Second, we analyse the same leisure walk data in relation to urban network analysis measures, such as space syntax variables like integration and normalised choice. We also investigate the impact of “perceptive greenness” on leisure walk, which is a measure of the amount of green elements that are directly seen by leisure walkers. This measure has been determined by analysing millions of Google Street View images along Singapore’s street network, categorising their content using automated deep learning algorithms. Finally, multi-variate spatial regression models are used to determine the importance of the defined features on the amount of leisure walks. The preliminary results show that there are positive associations with features such as transportation availability, land-use mix, perceived greenness and space syntax variables. The findings can help planners and designers understand which features promote more leisure walks, and inform the design of healthier, more resident-friendly environments that promote walking, and stimulate vibrancy.