Using Geographic Information and Point of Interest to Estimate Missing Second-Hand Housing Price of Residential Area in Urban Space

The real estate market including second-hand real estate market plays an important role in Chinese economy. However, it is not easy to acquire price of the each pieces of residential area in a city, and for instance the data acquired from Internet in our paper can only cover 56% of residential area in Nanjing. To this end, our paper proposed an model to fill the missing price data by using price and locations of second-hand real estates and Point of Interest (POI) information which were acquired from Internet. Our experiment was conducted in Nanjing and Chongqing, and demonstrates that our model is able to perform better than traditional Geographic Information System (GIS) method, such as Kriging interpolation, and general machine learning model, such as K-Nearest Neighbour (KNN). Also, our proposed model can be more interpretable than traditional methods, and able to reveal how the POI information can influence the second-hand real estate price. Our proposed model can help domain experts, e.g. city planners and economists, to better research the second-hand real estate market in the future.