Scholarlink

<p>Purpose: This paper investigated whether spatial big data mining of parcel-level building density and housing informality can be used to infer latent urban land market structures in data-scarce environments. Focusing on Kigali, the study examined whether observable spatial morphology can serve as a proxy for underlying land market segmentation in the absence of reliable transaction price data. Methodology: The analysis employed a comprehensive geospatial dataset covering 423,598 parcels across Kigali. Multiple scalable clustering algorithms (KMeans, MiniBatchKMeans, Fuzzy-KMeans, SOM-proxy, BIRCH, and Gaussian Mixture Models) were run and evaluated across ranging from 3 to 14 using Calinski–Harabasz, Davies–Bouldin and Silhouette indices. The optimal segmentation was selected based on statistical robustness, spatial coherence and interpretability. Findings: Results revealed four distinct and spatially clustered density–informality regimes: (1) an ultra-high-density formal core, (2) a moderate-density formal zone, (3) a consolidating informal settlement belt and (4) a low-density formal expansion periphery. These regimes are interpreted as proxies of underlying land market segmentation rather than direct observations of market structures. The configuration provides indicative evidence of differentiated spatial regimes that may reflect a dual urban land market structure within a formally registered tenure system. Segmentation arises not from tenure absence but from differences in development intensity and regulatory compliance, indicating that informality functions as an endogenous component of constrained land market dynamics. Originality: The paper advanced a novel spatial big data mining framework that bridges urban economic theory and high-resolution morphological analytics. By inferring dual land market structures from parcel-level spatial data, it contributed a replicable methodological alternative for housing market analysis in the Global South.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Spatial Big Data Mining of Building Density and Housing Informality: Inferring Dual Urban Land Market Structures in Kigali

  • 期刊论文

  • 发表日期：

    2026年6月2日

查看全文
  • Iyandemye Samuel,
  • Japhet Niyobuhungiro,
  • Edward Bbaale

摘要

Purpose: This paper investigated whether spatial big data mining of parcel-level building density and housing informality can be used to infer latent urban land market structures in data-scarce environments. Focusing on Kigali, the study examined whether observable spatial morphology can serve as a proxy for underlying land market segmentation in the absence of reliable transaction price data. Methodology: The analysis employed a comprehensive geospatial dataset covering 423,598 parcels across Kigali. Multiple scalable clustering algorithms (KMeans, MiniBatchKMeans, Fuzzy-KMeans, SOM-proxy, BIRCH, and Gaussian Mixture Models) were run and evaluated across ranging from 3 to 14 using Calinski–Harabasz, Davies–Bouldin and Silhouette indices. The optimal segmentation was selected based on statistical robustness, spatial coherence and interpretability. Findings: Results revealed four distinct and spatially clustered density–informality regimes: (1) an ultra-high-density formal core, (2) a moderate-density formal zone, (3) a consolidating informal settlement belt and (4) a low-density formal expansion periphery. These regimes are interpreted as proxies of underlying land market segmentation rather than direct observations of market structures. The configuration provides indicative evidence of differentiated spatial regimes that may reflect a dual urban land market structure within a formally registered tenure system. Segmentation arises not from tenure absence but from differences in development intensity and regulatory compliance, indicating that informality functions as an endogenous component of constrained land market dynamics. Originality: The paper advanced a novel spatial big data mining framework that bridges urban economic theory and high-resolution morphological analytics. By inferring dual land market structures from parcel-level spatial data, it contributed a replicable methodological alternative for housing market analysis in the Global South.