利用可解释性机器学习提取影响杭州市二手住宅价格的因子  

Using Interpretable Machine Learning to Extract Factors Affecting the Price of Second-Hand Houses in Hangzhou

作　　者：曹献之 陆激[2,3] 

机构地区：[1]浙江大学建筑工程学院 [2]浙江大学平衡建筑研究中心 [3]浙江大学建筑设计研究院有限公司

出　　处：《建筑与文化》2025年第2期226-228,共3页Architecture & Culture

摘　　要：地理人工智能近年来取得了飞速的发展,然而由于机器学习“黑箱”模型的天性导致其可解释性一直存在困境。文章以杭州市中心城区二手住宅价格为例,使用XGBoost模型将其与周边设施可达性指标和房产本身特征指标进行回归,并用SHAP模型解释各个变量的重要程度,最终量化了各个因子对于二手住宅价格的影响程度。结果发现,到最近地铁站的距离和房龄对二手房价格有最为显著的影响。此外,西湖周边和钱塘江东岸沿江一带表现出了很强的房价区位优势。Geographic artificial intelligence has achieved rapid development in recent years,but its interpretability has been difficult due to the nature of machine learning Black Box Models.This study examines the determinants of second-hand residential property prices in central Hangzhou.Utilizing the XGBoost Model,we regress property prices on the accessibility of surrounding facilities and intrinsic property characteristics.To elucidate the significance of each variable,the SHAP Model was employed,thereby quantifying the impact of each factor on property prices.The analysis revealed that proximity to the nearest metro station and the age of the property had the most significant effect on the price of second-hand houses.Additionally,properties located in the vicinity of West Lake and along the east bank of the Qiantang River demonstrated substantial locational advantages,reflected in higher property prices.

关 键 词：可解释性机器学习 XGBoost SHAP 二手住宅价格 杭州 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TU984.1[自动化与计算机技术—控制科学与工程] F299.23[建筑科学—城市规划与设计]