Developing a Strategic Crisis Management Model for Enhancing Stability of the Real Estate Sector: A Case Study of Chinese Cities

Abstract

The real estate bubble and economic crisis have significant impacts on Chinese real estate enterprises. This paper constructs a strategic crisis management model for improving the stability of the real estate industry based on a machine learning model. Based on the research data of cases in Chinese cities, the PSR model is introduced to construct a strategic crisis warning index system. It is proposed to use the LASSO feature selection method to conduct refined screening of feature indicators, and use it as the input layer of XGBoost. After optimizing the hyperparameters using the genetic algorithm, the model training and prediction are carried out. The results show that the LASSO-GA-XGBoost model has a higher warning accuracy rate for the stability crisis of the real estate industry compared to other models. The average ACC value and AUC value for early three-year crisis warning are 96.39% and 90.71%, respectively. Moreover, the prediction results have smaller errors compared to the actual results. Based on SHAP, the parameter features are further analyzed, and it is found that the importance ranking is the growth rate of real estate development investment and the rolling year-on-year income of land transfer fees.