Modeling urban land markets in data-scarce cities: a spatial big data mining approach to building density patterns in Kigali

Abstract

Purpose The purpose of this paper is to address the critical lack of traditional data in rapidly urbanizing, data-scarce cities by proposing a novel spatial big data mining framework that leverages building density as a reliable proxy for urban land market patterns. Design/methodology/approach This study used building density to infer urban land market patterns in Kigali, Rwanda. The core analysis confirmed significant spatial clustering (Moran’s I = 0.9780) and multi-metric validation of five clustering algorithms selected the k-means model (k = 5) for robust urban segmentation. Findings The clustering delineated five distinct housing density zones, confirming a clear spatial gradient consistent with the classical bid-rent theory and monocentric city model. The high-density core (density: 0.34) comprises 9.93% of the land area, while extensive low-density zones dominate the periphery, empirically validating the applicability of traditional urban economic models in this data-scarce African context. Practical implications This study provides urban planners and policymakers with an evidence-based map of land market pressure. This granular segmentation enables targeted land-use planning, optimized infrastructure investment and the development of equitable policies for managing urban growth and densification in the future. Originality/value This study used building footprints density to infer land market patterns in Kigali, offering replicable methodology for data-driven spatial analysis in the Global South.