Performance Analysis of 15-Minute City Zones Through Spatial and Machine Learning Techniques

Abstract

This paper presents a methodology for evaluating pedestrian accessibility within the 15-minute city framework, using the Küçükçekmece district of Istanbul as a case study. The research integrates spatial clustering, GIS-based accessibility analysis, slope-adjusted walking speeds derived from digital elevation models, and categorized point-of-interest (POI) data to assess access to urban services within 15-, 20-, and 25-minute walking thresholds. Building footprints were clustered using the Mean Shift machine learning algorithm, identifying 42 urban clusters that served as focal points for accessibility analysis. The results indicate that 71.5% of the district is accessible within a 15-minute walk, increasing to 85.5% and 91.6% for 20- and 25-minute intervals, respectively. However, significant spatial inequalities persist, particularly in peripheral neighbourhoods, newly developed areas, and locations with poor internal pedestrian connectivity. Cultural facilities were found to be the least accessible category of urban services, while daily necessities and commercial functions were relatively well distributed. The study demonstrates how urban form, topography, and the spatial distribution of services influence walkability and accessibility, providing a transferable framework for evaluating proximity-based urban planning and supporting more equitable, resilient, and pedestrian-oriented city development.