集成PSO-NN模型在房价预测中的应用研究  

Research on Application of Integrated PSO-NN Model in Housing Price Prediction

作　　者：段永辉[1] 高绅 郭一斌[2] 王翔[2] DUAN Yong-hui;GAO Shen;GUO Yi-bin;WANG Xiang(School of Civil Engineering,Henan University of Technology,Zhengzhou Henan 450001,China;School of Civil Engineering,Zhengzhou University of Aeronautics,Henan Zhengzhou 450015,China)

机构地区：[1]河南工业大学土木工程学院,河南郑州450001 [2]郑州航空工业管理学院土木工程学院,河南郑州450015

出　　处：《计算机仿真》2022年第10期476-480,518,共6页Computer Simulation

基　　金：国家自然科学基金(81973791);河南省高等学校重点科研项目计划(13B520878,14A630029)。

摘　　要：针对商用住宅价格预测问题,提出了一种基于集成学习的多粒子群优化神经网络预测模型(PSO-NN)。基于文献回顾法,建立商用住宅价格指标体系;其次,提出一种利用粒子群算法(PSO)优化神经网络(NN)权重的PSO-NN模型;进一步提出一种基于bagging集成思想的PSO-NN模型,将新模型应用于郑州市住宅价格预测问题。实验结果表明,集成PSO-NN模型相对于支持向量机、线性回归等传统预测模型,在均方误差、平均绝对误差两项指标上分别提升了26.14%和27.61%;通过可信度分析,集成模型预测结果具有较高的可信度,进一步说明该模型在住宅价格预测中具有较大的优势。结果证明,针对商用住宅价格预测问题,基于bagging的集成学习思想可以进一步提升改进神经网络模型的精度。Aiming at the problem of commercial residential price prediction, a multi-particle swarm optimization neural network prediction model(PSO-NN) based on ensemble learning is proposed. First, establish a commercial residential price index system based on the literature review method;Second, propose a PSO-NN model that uses particle swarm optimization(PSO) to optimize neural network(NN) weights;Third, further propose a bagging integrated idea PSO-NN model;Finally, apply the new model to the residential price prediction problem in Zhengzhou. The experimental results show that, compared with traditional prediction models such as support vector machines and linear regression, the integrated PSO-NN model has increased by 26.14% and 27.61% respectively in the two indicators of Mean square error andMean absolute error;Through credibility analysis, the prediction results of the integrated model have high credibility, which further shows that the model has a greater advantage in residential price prediction. The results prove that for the problem of commercial residential price prediction, the integrated learning idea based on bagging can further improve the accuracy of the improved neural network model.

关 键 词：商用住宅 价格预测 粒子群优化算法 集成学习 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]