Urban real estate market early warning based on support vector machine: A case study of Beijing

Abstract

Based on a multi-class support vector machine, an urban real estate early warning model is constructed for the Beijing real estate market. The initial indicator system is established based on the historical development of Beijing’s real estate market and the selection of real estate early warning indicators. Early warning index data for Beijing from 2000 to 2018 are selected, and the leading index is selected by a time difference correlation analysis as the warning index to be used for further implementation. The model is found to have a good early warning judgment performance, and demonstrates generalization ability. The model analyzes the real estate market from the aspects of land supply, credit scale, housing supply structure, restriction of speculation, and strengthening of transparency of real estate information. It predicts that the real estate market in Beijing will run smoothly in 2019. Based on the model’s findings, the paper proposes policy recommendations to promote the healthy operation of China’s real estate market.