Apartment Rent Price Estimation from the Image Perspective: A case study in Beijing

Abstract

In China, the real-estate market has been gradually attracting the attention of people owing to soaring housing prices. The rental housing market played a vital role of the real-estate economy in Chinese cities. However, the market lacks reasonable base pricing because of the unavailability of public data and standard policies. Moreover, the existing valuation of houses primarily relies on the description given by people through textual information; thus, the tenant does not get an intuitive understanding of the actual condition of the house. The introduction of the Internet has enabled the development of websites that provide a platform to renters and buyers; however, the rental prices on the Internet are determined by the landlord or the agent, and these valuations are not based on standard pricing. The pivotal factors that can affect the valuation of a house are the conditions of the house and the facilities in it. To achieve a more intuitive estimation of rental prices, this study develops a machine-learning-based house-rent-evaluation model that uses a large number of indoor pictures, together with information from open data. We show that this model is feasible, and it also offers an improved rent-evaluation accuracy.