Spatial examination of house prices via cluster analysis

Purpose In Ankara, house prices exhibit significant variation across different locations, a trend closely tied to speculative behaviors in the Turkish housing market. Many neighborhoods remain unaffordable, even for households within the highest income quintile. This raises critical questions: To what extent are these pronounced price differences attributable to the specific attributes of the housing stock? Can Ankara’s housing market be segmented into homogeneous clusters based on housing characteristics and price levels? What are the primary factors influencing housing prices, and what role does speculation play in shaping these dynamics? This study aims to address these questions through the application of various statistical analyses. Design/methodology/approach This study examines housing market differentiation in Ankara by applying cluster analysis, discriminant analysis and hedonic price analysis, using data from 12,460 dwelling units across 325 neighborhoods within eight central districts of the city. Findings The findings of this study indicate that speculation can be a significant determinant of house prices, as evidenced by the considerable price variations observed even among properties with similar physical characteristics. Originality/value Cluster analysis, a commonly used statistical method in the segmentation of housing submarkets, plays a central role in this research. However, unlike prior studies, this paper tries to explore the existence of homogeneous clusters within Ankara’s housing markets in the context of housing affordability and housing speculation via different statistical techniques.