基于人工神经网络的中国房地产市场预警及实证研究

发布时间：2018-03-15 12:15

本文选题：房地产市场预警　切入点：人工神经网络　出处：《东北财经大学》2013年硕士论文　论文类型：学位论文

【摘要】：随着住房制度改革和城市化进程的加快,我国房地产业迅速发展,对我国国民经济的影响也越来越大。房地产业的蓬勃发展对改善居民生活、拉动经济增长发挥了重要作用。然而,房地产投资额的大幅增长也使得房地产业出现了一系列问题。近年来,我国房价逐年上涨,市场风险逐步加大,关于房地产泡沫的讨论越来越激烈。因此,建立城市房地产预警系统,对我国房地产市场的发展状况进行准确的判断和监控,对确保房地产业持续健康发展意义重大。房地产预警是经济预警的组成部分,本文首先总结了国内外关于经济预警的研究动态,然后分析了我国房地产预警领域的研究现状及存在的主要问题。通过对现有预警方法的优缺点进行对比分析,鉴于人工神经网络具有非线性、容错性、操作简便等优点,本文决定采用人工神经网络预警方法构建本文的房地产预警模型。在大量阅读整理相关文献的基础上,从房地产业发展速度、房地产业与国民经济的协调关系、房地产业内部协调关系三个方面选择了最常用、代表性较强的指标构建本文的房地产预警指标体系。基于人工神经网络的房地产预警模型的构建思路如下：首先,根据研究对象特点选择警情指标,判断以往年份的房地产市场警情,本文采用的是归一化方法。然后,确定警兆指标,警兆指标应当对警情指标的领先性,还要考虑到数据的可获得性等因素。最后,采用Matlab软件编写了人工神经网络的训练程序对样本数据进行训练,得出我国房地产市场的预警模型。该预警模型经过检验后即可运用于预警实践。为了对我国房地产市场整体运行状况有一个更为全面的了解,本文按照经济发展程度的差别分别选取了北京、重庆、威海作为我国一二三线城市的代表城市进行实证研究,通过数据收集,分别建立了人工神经网络的训练样本,并对预警结果的准确性进行了检验,然后将2012年房地产市场数据处理后得出的警兆指标输入预警模型对下一年的房地产警情进行判断,预警结果显示北京市和威海市2013年房地产市场将处于“热”的警情状态下,2013年重庆市房地产市场将处于“正常”的警情状态下。在得出预警结论的基础上,文章最后简单分析了出现这种结果的原因,政府可从规范土地管理制度,调节土地供应、控制房地产信贷规模、调整住房供应结构、限制房地产市场投机和炒作行为、增加房地产市场信息透明度等方面加强对房地产市场的管理。由于我国房地产预警研究还不够成熟,多数预警方法还仅仅停留在理论层面,距离实际应用还有较多问题,尽管人工神经网络预警方法有其自身优越性,是较为先进的科学方法,但我国房地产数据不足、关于房地产预警指标体系研究不够成熟等因素也会使得该方法的运用受到较大局限。在后续的研究中还需要在这些方面有所加强。
[Abstract]:With the reform of housing system and the acceleration of urbanization, the real estate industry in our country is developing rapidly, and the influence on our national economy is becoming more and more great. Promoting economic growth has played an important role. However, the substantial increase in real estate investment has also caused a series of problems in the real estate industry. In recent years, housing prices in China have increased year by year, and the market risks have gradually increased. The discussion on the real estate bubble is becoming more and more intense. Therefore, it is of great significance to establish the urban real estate early warning system to accurately judge and monitor the development of the real estate market in our country, which is of great significance to ensure the sustained and healthy development of the real estate industry. Real estate early warning is an integral part of economic early warning. This paper first summarizes the research trends of economic early warning both at home and abroad. Then it analyzes the current research situation and main problems in the field of real estate early warning in China. By comparing and analyzing the advantages and disadvantages of the existing early warning methods, in view of the advantages of artificial neural network, such as nonlinear, fault-tolerant, easy to operate, etc. This article decides to use the artificial neural network early warning method to construct the real estate early warning model of this paper. On the basis of a large number of reading and sorting relevant literature, from the real estate industry development speed, real estate industry and national economy coordination relations, Three aspects of the coordination relationship within the real estate industry choose the most commonly used, more representative indicators to build the real estate warning index system. The real estate warning model based on artificial neural network is constructed as follows: first, According to the characteristics of the object of study, select the alarm index, judge the real estate market alarm situation in the past years, this paper adopts a normalized method. Then, determine the warning indicators, warning indicators should be the lead to the warning indicators, Finally, the training program of artificial neural network is compiled by Matlab software to train the sample data. The early warning model of China's real estate market is obtained, which can be used in early warning practice after being tested. In order to have a more comprehensive understanding of the overall operation of China's real estate market, this paper selects Beijing, Chongqing and Weihai as the representative cities of the 123 tier cities in China according to the difference of economic development. Through data collection, the training samples of artificial neural network are established, and the accuracy of early warning results is tested. Then input the warning indicators of the 2012 real estate market data processing into the early warning model to judge the real estate police situation in the next year. The warning results show that the real estate market in Beijing and Weihai will be in a "hot" state on 2013, and that in Chongqing on 2013, the real estate market will be in a "normal" state of warning. On the basis of the conclusion of early warning, the paper analyzes the reason of the result. The government can regulate the land supply, control the scale of real estate credit, adjust the structure of housing supply. To restrict speculation and speculation in the real estate market and to increase the transparency of information in the real estate market, and so on, to strengthen the management of the real estate market. Because the research of real estate early warning in our country is not mature enough, most of the early warning methods are still only in the theoretical level, and there are still many problems from the practical application, although the artificial neural network early warning method has its own superiority. Is a more advanced scientific method, but the real estate data in our country are insufficient. The application of this method will also be limited by some factors, such as the immaturity of the research on the early warning index system of real estate, and it is necessary to strengthen these aspects in the follow-up research.
【学位授予单位】：东北财经大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP183;F299.23

【参考文献】