XGBoost算法在二手房价格预测中的应用  

The Application of XGBoost Algorithm in Second-Hand House Price Prediction

作　　者：王美 朱珊艳 陈鹏蕾 武业文[1] 

机构地区：[1]南京信息工程大学数学与统计学院,江苏 南京 [2]浙江科技学院理学院,浙江 杭州

出　　处：《运筹与模糊学》2023年第2期734-744,共11页Operations Research and Fuzziology

摘　　要：随着近些年社会经济的快速发展,房地产的开发也如火如荼展开,二手房市场也得到迅猛发展,在如今存量房时代,二手房交易成为房地产市场的重要部分,二手房的价格是诸多购房者的关注点,因此对二手房价格预测是有必要的。本文围绕XGBoost算法学习,借助网络爬虫技术,从链家网站采集了杭州九堡1500条在售二手房信息,将数据清洗并提取特征后,在Shiny平台进行展示,并以2023年2月江干区二手房均价34,381元/平米为分界线,将该地二手房分为高、低价格共两类。本文就影响二手房房价的因素进行深入研究,进而对房价进行分类预测。通过对影响二手房价格的特征因素提取并排序,结果显示第一重要的特征是房屋面积,其次是二手房屋关注的人数及发布时间、建筑结构、房间数量、房屋装修情况这四个特征,而房本时间、客厅和餐厅数量是重要性最弱的特征。本文基于XGBoost算法对房价预测,结果分类效果较为理想,说明算法的应用性较好,同时为后续我国二手房价格预测或其他问题的预测扩充探索的道路。With the rapid socio-economic development in recent years, real estate development has been in full swing and the secondary housing market has also developed rapidly. Nowadays, in the era of inventory, second-hand house transactions have become an important part of the real estate market, and the price of second-hand houses is a concern for many home buyers, so it is necessary to predict the price of second-hand houses. In this paper, around XGBoost algorithm, with the help of web crawler technology, 1500 second-hand houses for sale in Jiubao, Hangzhou are collected from the website of Chain Home, the data are cleaned and features are extracted and displayed in Shiny platform, and the average price of second-hand houses in Jianggan District in February 2023 is 34,381 Yuan every square meter as the dividing line, and the second-hand houses in the area are divided into two categories of high and low prices in total. This paper conducts an in-depth study on the factors affecting the price of second-hand houses, and then categorizes and predicts the price of houses. By extracting and ranking the feature factors affecting the price of second-hand houses, the results show that the first important feature is the house area, followed by the four features of the number of people concerned about second-hand houses and the release time, the building structure, the number of rooms, and the house decora-tion, while the time of the house book and the number of living and dining rooms are the features with the weakest importance. This paper is based on the XGBoost algorithm for house price prediction, and the results of the classification effect is more satisfactory, which indicates that the algorithm has better applicability, and also expands the path of exploration for the subsequent prediction of second-hand house price or other problems in China.

关 键 词：XGBoost 网络爬虫 分类预测 二手房 

分 类 号：F29[经济管理—国民经济]