基于麻雀搜索优化SVR模型的房地产价格研究

doi:10.16180/j.cnki.issn1007-7820.2024.01.001

摘要/Abstract

摘要：

为解决传统经济指标作为房价影响因素的数据获取滞后性问题以及机器学习模型在预测房价时存在的参数选取不确定性问题,文中以网络搜索数据作为房价指数解释变量,采用麻雀搜索算法(Sparrow Search Algorithm, SSA)建立SSA-SVR(Support Vector Regression)模型对SVR的惩罚因子C和RBF(Radical Basic Function)核函数的参数g进行优化。将SSA-SVR模型与PSO(Particle Swarm Optimization)-SVR、GA(Genetic Algorithm)-SVR、WOA(Whale Optimization Algorithm)-SVR、GS(Grid Search)-SVR以及基准SVR进行对比,SSA-SVR的相关系数(0.99)、均方根误差(6.71)、平均绝对误差(5.24)、均方误差(45.13)以及平均绝对百分比误差(0.26%)均优于其他5种模型。结果表明,麻雀搜索算法优化的SVR模型在房价预测方面具有更好的全局寻优能力,可以提高模型的预测准确度和预测能力。

关键词: 麻雀搜索算法, 优化算法, SVR模型, 数据滞后性, 参数不确定性, 网络搜索数据, 房地产价格指数, 房价预测

Abstract:

In order to solve the data acquisition lag problem of traditional economic indicators as an influencing factor of housing prices, and the uncertainty of parameter selection in the machine learning model when predicting housing prices, the network search data is used as the explanatory variable of the house price index, and Sparrow Search Algorithm(SSA) is used to establish the SSA-SVR(Support Vector Regression)model to optimize the penalty factor C of SVR and the parameter g of the RBF(Radical Basic Function) kernel function in this study. Comparison among the established SSA-SVR model with PSO(Particle Swarm Optimization)-SVR, GA(Genetic Algorithm)-SVR,WOA(Whale Optimization Algorithm)-SVR, GS(Grid Search)-SVR and benchmark SVR show that the correlation coefficient of SSA-SVR(0.99), root mean square error(6.71), mean absolute error(5.24), mean square error(45.13) and mean absolute percentage error(0.26%) are better than those of the other five models. The results show that the SVR model optimized by the sparrow search algorithm has better global optimization ability in housing price prediction, which can improve the prediction accuracy and prediction ability of the model.

Key words: sparrow search algorithm, optimization algorithm, SVR model, data lag, parameter uncertainty, network search data, real estate price index, house price forecast

中图分类号:

TP39

兰瑞杰,孟维高,耿进强. 基于麻雀搜索优化SVR模型的房地产价格研究[J]. 电子科技, 2024, 37(1): 1-8.

LAN Ruijie,MENG Weigao,GENG Jinqiang. Research on Real Estate Price Index Based on Sparrow Search Optimization SVR Model[J]. Electronic Science and Technology, 2024, 37(1): 1-8.

图/表 8

图1

表1

图2

图3

表2

图4

图5

表3

参考文献 20

[1]	孙毅, 吕本富, 陈航, 等. 基于网络搜索行为的消费者信心指数构建及应用研究[J]. 管理评论, 2014, 26(10):117-125.
	Sun Yi, Lü Benfu, Chen Hang, et al. Research on the establishment and application of the consumer confidence index based on network search behavior[J]. Management Review, 2014, 26(10):117-125.
[2]	白丽娟, 闫相斌, 金家华. 基于搜索关键词关注度的商品房价格指数预测[J]. 工程管理科技前沿, 2015, 34(4):65-70.
	Bai Lijuan, Yan Xiangbin, Jin Jiahua. Forecast the com-mercial housing price index based on search keywords attention[J]. Frontiers of Science and Technology of Engineering Management, 2015, 34(4):65-70.
[3]	王博永, 杨欣. 基于网络搜索的房地产政策调控效果研究[J]. 管理评论, 2014, 26(9):78-88.
	Wang Boyong, Yang Xin. A study of real estate market macro-control policies' effectiveness based on web search data[J]. Management Review, 2014, 26(9):78-88.
[4]	张爱琳, 白丽娜. 基于GM(1,1)模型的包头市房地产价格预测[J]. 建筑经济, 2020, 41(S2):266-269.
	Zhang Ailin, Bai Lina. Price forecast of Baotou real estate based on GM(1,1) model[J]. Construction Economy 2020, 41(S2):266-269.
[5]	肖卫国, 陈宇, 尹智超. 房地产价格对中国宏观经济的影响研究[J]. 统计与决策, 2016(2):124-126.
	Xiao Weiguo, Chen Yu, Yin Zhichao. Research on the i-mpact of real estate prices on China's macroeconomic[J]. Statistics and Decisions, 2016(2):124-126.
[6]	孙爱荣, 程亚鹏. 基于灰色理论和BP神经网络的房地产价格指数预测[J]. 企业经济, 2010(4):124-126.
	Sun Airong, Cheng Yapeng. Real estate price index prediction based on grey theory and BP neural network[J]. Enterprise Economics, 2010(4):124-126.
[7]	刘晓君, 胡升凯, 迟依涵. 基于VAR-GM(1.1)-SVR模型的房地产价格预测研究[J]. 数学的实践与认识, 2021, 51(1):1-12.
	Liu Xiaojun, Hu Shengkai, Chi Yihan. Research on real estate price forecast based on VAR-GM(1.1)-SVR model[J]. Mathematics in Practice and Theory, 2021, 51(1):1-12.
[8]	董倩, 孙娜娜, 李伟. 基于网络搜索数据的房地产价格预测[J]. 统计研究, 2014, 31(10):81-88.
	Dong Qian, Sun Nana, Li Wei. Real estate price predict-ion based on web search data[J]. Statistical Research, 2014, 31(10):81-88.
[9]	唐晓彬, 张瑞, 刘立新. 基于蝙蝠算法SVR模型的北京市二手房价预测研究[J]. 统计研究, 2018, 35(11):71-81.
	Tang Xiaobin, Zhang Rui, Liu Lixin. Research on forecast of second-hand house price in Beijing based on SVR model of bat algorithm[J]. Statistical Research, 2018, 35(11):71-81.
[10]	Wang X B, Wen J H, Zhang Y H, et al. Real estate price forecasting based on SVM optimized by PSO[J]. Optik International Journal for Light and Electron Optics, 2014, 125(3):1439-1443. doi: 10.1016/j.ijleo.2013.09.017
[11]	赵怡爽. 房地产价格影响因素分析及预测[J]. 统计与决策, 2014(13):107-110.
	Zhao Yishuang. Analysis and forecast of influencing factors of real estate prices[J]. Statistics & Decision, 2014(13):107-110.
[12]	战松, 孙川. 基于VAR模型的沈阳市房价影响因素分析[J]. 沈阳建筑大学学报(社会科学版), 2022, 24(2):167-172.
	Zhan Song, Sun Chuan. Influencing factors analysis of housing price in Shenyang based on VAR model[J]. Journal of Shenyang Jianzhu University(Social Science), 2022, 24(2):167-172.
[13]	綦文彬, 侯宗润, 李贵熙. 基于GM(1,1)模型及BP神经网络的四城市房价分析与预测[J]. 电脑知识与技术, 2018, 14(5):179-185.
	Qi Wenbin, Hou Zongrun, Li Guixi. Analysis and predi-ction for housing price of four cities based on GM(1,1) model and neural net methods[J]. Computer Knowledge and Technology, 2018, 14(5):179-185.
[14]	刘晓君, 胡升凯, 迟依涵. 基于VAR-GM(1.1)-SVR模型的房地产价格预测研究[J]. 数学的实践与认识, 2021, 51(1):1-12.
	Liu Xiaojun, Hu Shengkai, Chi Yihan. Research on real estate price forecast based on VAR-GM(1.1)-SVR model[J]. Mathematics in Practice and Theory, 2021, 51(1):1-12.
[15]	马勇, 吴雪妍. 银行信贷如何影响房价[J]. 金融评论, 2018, 10(3):1-22,122.
	Ma Yong, Wu Xueyan. How does bank credit affect real estate price-evidence from 30 Chinese provinces[J]. Chinese Review of Financial Studies, 2018, 10(3):1-22,122.
[16]	杜美君, 张伟, 谢亚莲. 基于粒子群算法的PID控制器参数优化[J]. 电子科技, 2019, 32(6):7-11.
	Du Meijun, Zhang Wei, Xie Yalian. Particle swarm optimization based PID controller parameter optimization[J]. Electronic Science and Technology, 2019, 32(6):7-11.
[17]	张锦, 陈林, 赖祖龙. 改进遗传算法优化灰色神经网络隧道变形预测[J]. 测绘科学, 2021, 46(2):55-61,77.
	Zhang Jin, Chen Lin, Lai Zulong. Tunnel deformation prediction based on grey neural network with improved genetic algorithm[J]. Science of Surveying and Mapping, 2021, 46(2):55-61,77.
[18]	沙金霞. 改进鲸鱼算法在多目标水资源优化配置中的应用[J]. 水利水电技术, 2018, 49(4):18-26.
	Sha Jinxia. Application of ameliorative whale optimization algorithm to optimal allocation of multi-objective water resources[J]. Water Resources and Hydropower Engineering, 2018, 49(4):18-26.
[19]	张文静, 刘茵, 张孟, 等. 面向VR建模的网格简化算法优化研究[J]. 电子测试, 2021(20):75-76.
	Zhang Wenjing, Liu Yin, Zhang Meng, et al. Optimization of mesh simplification algorithm for VR modeling[J]. Electronic Test, 2021(20):75-76.
[20]	刘琴, 黄襄念. 一种改进的自适应GA-SVM参数选择研究[J]. 电子科技, 2010, 23(5):99-101.
	Liu Qin, Huang Xiangnian. Parameter selection of support vector machines based on an improved adaptive genetic algorithm[J]. Electronic Science and Technology, 2010, 23(5):99-101.

因素类别	初始关键词
财政因素	房地产交易税房地产税房地产营业税房屋买卖税商品房税
财政因素	土地增值税物业税契税
金融因素	按揭按揭买房存款基准利率存款利率存款准备金率贷款利率贷款利率下调房贷房屋贷款购房贷款加息降息
金融因素	买房贷款期房贷款
政策因素	天津购房政策天津落户天津限购限购政策房地产调控国八条
政策因素	国六条国十条国十一条国四条国五条人才引进
可替代性因素	保障房保障性住房限价房公租房经济适用房廉租房天津二手房
供给市场因素	征地补偿容积率土地出让土地储备土地使用权房地产投资
供给市场因素	需求方因素:天津买房天津户口天津学区房中国指数研究院数据库天津房产网安居客链家房天下工资水平物价 CPI
需求市场因素	天津买房天津户口天津学区房中国指数研究院数据库天津房产网安居客链家房天下工资水平物价 CPI

样本	真实值	预测值	样本	真实值	预测值	样本	真实值	预测值
1	1 952	1 943	13	1 858	1 865	25	1 925	1 930
2	2 197	2 194	14	1 887	1 888	26	2 196	2 196
3	2 155	2 156	15	2 157	2 167	27	2 169	2 159
4	1 864	1 866	16	2 140	2 137	28	2 152	2 158
5	2 199	2 197	17	2 196	2 194	29	2 026	2 026
6	1 899	1 910	18	1 920	1 924	30	1 896	1 902
7	1 943	1 939	19	1 887	1 886	31	1 988	1 987
8	2 158	2 159	20	1 868	1 857	32	2 092	2 089
9	1 890	1 900	21	2 157	2 164	33	2 195	2 194
10	1 857	1 861	22	1 958	1 951	34	2 156	2 158
11	2 188	2 194	23	2 171	2 171	35	1 881	1 868
12	1 855	1 861	24	1 852	1 853	36	1 942	1 932

	SSA- SVR	PSO- SVR	GA- SVR	WOA- SVR	GS- SVR	SVR
MAE	5.24	13.11	7.81	6.21	36.12	31.42
MSE	45.13	199.06	205.63	62.41	1 783.37	1 246.87
MAPE	0.26%	0.65%	0.39%	0.33%	12.90%	11.42%
RMSE	6.71	16.25	14.34	7.92	42.23	35.31
r	0.99	0.99	0.99	0.99	0.85	0.93