Gaps and overlaps of urban housing sub-market: Hard clustering and fuzzy clustering approaches

It has long been argued that the housing market is spatially subdivided within an urban area. The argument has important implications for explaining how the housing market works and describing the distinctiveness of each housing submarkets, having determined, a priori, its segmentation. The most commonly used method for identifying housing submarkets is based on cluster analysis, although hedonic analysis has been extensively used. The hedonic analysis is used to derive dimensionality of the housing market by estimating what attributes are significant factors influencing housing price. Those attributes or variables can then be used for cluster analysis. The paper proposes an analysis of the real estate market in San Cristoforo, Catania, trying to integrate two different clustering analysis approaches to defining its possible submarkets articulation. The first one is a hard clustering approach using the K-means method and hypothesizing different numbers of clusters. The second one can be considered a verification of the previous results: a fuzzy algorithm is applied to obtain the fuzzy set membership degree of each data point to housing submarkets defined within the examined urban area. The comparison between the results coming from the two different approaches suggests some reflections about the use of these powerful techniques for integrating the knowledge of the complex and multi-layered real estate markets in the urban recovery policies.