Abstract
Accurate pricing of the property market is necessary to ensure effective and efficient decision making. Property price is typically modelled using the hedonic price model (HPM). This approach was found to exhibit aggregation bias due to its assumption that the coefficient estimate is constant and fails to consider variation in location. The aggregation bias is minimized by segmenting the property market into submarkets that are distinctly homogeneous within their submarket and heterogeneous across other submarkets. Although such segmentation was found to improve the prediction accuracy of HPM, there appear to be conflicting findings regarding what constitutes a submarket and how the submarkets are to be driven. This paper therefore reviews relevant literature on the subject matter. It was found that, initially, submarkets were delineated based on a priori classification of the property market into predefined boundaries. The method was challenged to be arbitrary and an empirically statistical data-driven property submarket classification was advocated. Based on the review, there is no consensus on the superiority of either of the methods over the another; a combination of the two methods can serve as a means of validating the effectiveness of property segmentation procedures for more accurate property price prediction.