Housing Market Segmentation Based on Apartment Descriptions

Abstract

Market segmentation is the process of breaking down diverse markets into homogenous groupings that have comparable demands, interests, and/or behaviours. Housing market, like any other market, has its own challenges when it comes to market segmentation, since the segmentation process depends not only on the product (houses or apartments) but also on the market player (owners or real-estate agents).

This paper aims to show how Natural language processing (NLP) can be used to determine the segments of the housing market in Yerevan by using the unstructured data (apartment descriptions) scrapped from a real-estate website. The collected textual data represents not only the descriptions of the apartments but also gives an idea of who wrote that text. The applied NLP model shows how certain behavioral patterns of market players can be expressed through textual data and how those patterns can affect market segmentation. That means the segmentation of the market using unstructured data represents not only product-related, but also psychographic and geographic picture of the customers (here – sellers) and their apartments.