With the increasing application of machine learning technologies in the real estate sector, traditional models face challenges in terms of prediction accuracy and capturing nonlinear relationships. This study aims to integrate classical statistical methods with advanced machine learning algorithms to thoroughly explore the core factors influencing housing prices in Guangzhou and achieve high-precision regional price predictions. By employing statistical hypothesis testing to preliminarily screen factors, the significance of their relationships with housing prices is verified. On this basis, an XGBoost learning model is constructed to perform modeling and prediction, leveraging its strong capabilities in handling nonlinear relationships, interaction effects, and missing data. The study identifies the number of new houses, GDP per capita, the number of subway stations, and the number of large supermarkets as key factors verified to have a significant positive impact on housing prices in Guangzhou. Housing prices in Guangzhou's core districts have taken the lead in recovery, while development zones have shown steady growth with the greatest potential. In contrast, the peripheral suburban areas continue to face downward pressure.
Junyang Liu (Wed,) studied this question.