摘要
为了更加精准地预测二手房价格,该文以2019年深圳市二手房的真实交易数据为研究对象,利用线性回归模型、随机森林模型和XGBoost模型并加以POI计算来预测二手房价格.首先,对数据集进行清洗并可视化展示.其次,运用百度地图进行POI处理扩充数据集,使得数据集接近现实情况.接着,按照数据特征对房价影响的重要程度进行了排序,选取重要的特征来训练模型.最后,通过数值结果分析,XGBoost模型对二手房的房价评估效果最好,尤其是经过POI处理的数据集和XGBoost模型的这种组合,对于深圳市的二手房价格具有极好的预测效果.
In order to predict the second-hand house prices more accurately,the real transaction data of second-hand houses in Shenzhen in 2019 is taken as the research object,using Linear Regression,Random Forest,XGBoost and add POI calculation to predict the price.Firstly,the dataset was cleaned and visualized.Then,the dataset was expanded by using Baidu map for POI processing to make it close to the reality.Next,the dataset features were ranked according to their importance in influencing house prices,and the important features were selected to train the models.Finally,the experimental data showed that XGBoost model was the best for predicting second-hand house prices,especially the combination of the POI dataset and XGBoost model had an excellent prediction performance for the second-hand house prices in Shenzhen.
作者
胡晓伟
马春梅
孔祥山
李凤银
HU Xiaowei;MA Chunmei;KONG Xiangshan;LI Fengyin(School of Computer Science,Qufu Normal University,276826,Rizhao,Shandong,PRC)
出处
《曲阜师范大学学报（自然科学版）》
CAS
2022年第1期57-65,共9页
Journal of Qufu Normal University(Natural Science)
基金
国家自然科学基金(62072273)
山东省重大基础研究(ZR201906140028)。
