Segmentation and Valuation in  Stockholm Housing Market : Spatial Continuous and Discontinuous Submarkets Evaluating by Hedonic Price Model and XGBoost Model

Sammanfattning: The housing market segmentation could provide a reference for more targeted policymaking and investment strategies. Although there have been many studies, there are no consistent submarkets delineating methods because of a lack of theoretical support and subjective evaluation. In this paper, two market segmentation methods are introduced. The continuous spatial segmentation divides properties into submarkets according to their coordinates, while the discontinuous spatial segmentation creates submarkets according to the variable having the most significant impact on the price index, which is the construction year of properties. Two valuation methods, the hedonic price model and the XGBoost regression model, are applied to evaluate the overall Stockholm housing markets and the created. The results proved that both market segmentation methods could improve the valuation prediction accuracy compared to the valuation under the overall Stockholm housing market. The non-spatial continuous market segmentation approach delivers more improvement in valuation accuracy but also has greater volatility. As for the two valuation models, no single valuation method can be absolutely advantageous in any market context.