Research on Second-Hand Housing Prices in Guangzhou Based on CHAID Algorithm and POI Data

Studying the spatial characteristics that affect housing prices helps to understand the spatial distribution pattern of housing prices. Traditional hedonic price models and geographically weighted regression models lack flexibility in solving nonlinear problems, while artificial neural networks can hardly explain the influence mechanism of different spatial characteristics on housing prices. This paper takes the second-hand ordinary residential districts in Guangzhou with a construction age of 5–10 years as the research object, and aims to identify the most important spatial characteristics that affect their prices. In view of the defects of above traditional methods, firstly, this article uses the Geographic Information System (GIS) technology in Baidu Maps to collect data. Then this study uses the fast and interpretable Chi-square Automatic Interaction Detection (CHAID) algorithm to explore the relationship between residential spatial characteristics and housing prices from the perspective of market segmentation. The research results show that the number of subway stations around the community, the distance from the Central Business District (CBD) and the number of surrounding hospitals are the key factors affecting housing prices, while the distance to the nearest subway station and the distance from surrounding hospitals are secondary factors. This research provides necessary references for the government’s real estate market regulation and urban planning. Meanwhile, the innovative application of CHAID algorithm also enriches the methodology of real estate market analysis.