存量房批量评估系统的研究与实现

发布时间：2018-07-31 07:25

【摘要】：由于单宗评估方法在房地产价值评估过程中存在评估标准不一、成本过大等诸多问题，进而导致一定程度上税收流失、社会不公等现象的产生，许多欧美发达国家多年前就开始采用计算机辅助的批量评估方法。计算机辅助批量评估（Computer-assisted Mass Appraisal，CAMA），即在原有成本法、市场比较法和收益法评估原理的基础上，结合计算机技术和统计分析技术，对大批量特定类型的物业的市场价值进行快速、高效地评估的一种方法。如今在欧美发达国家，批量评估已经逐步取代了单宗评估，广泛地应用于不动产的税基评估中。与之相比，我国的批量评估技术还处于应用起步阶段。得益于2009年下半年财政部和税务总局在各个地区开始进行的存量房批量评估试点和推广工作，目前我国大多数地区已经初步建立了CAMA系统，并已于2012年7月1日起正式对纳税人申报的存量房交易价格进行全面的评估。但由于各国不动产税基评估体系的设计与本国的经济结构、政府职能部门的设置、税制发展改革、法律体系等都有着密切关系，我国的CAMA系统建设不能照搬照抄国外的理论，特别是批量评估的核心——自动评估模型（Automated Valuation Model，AVM），，应考虑我国的实际情况，对国外成熟的批量评估技术进行本地化改进。本文首先详细介绍了批量评估技术的概念与方法原理，分析了国内外的发展情况，并系统总结了实施批量评估的具体步骤。实施CAMA系统有两个非常重要的关键点，一是构建数据全面、真实、准确的房地产基础信息数据库；二是构建一个合适特定物业类型的自动评估模型。本文对以上两个关键点的实施进行了相关论述，并着重介绍了自动评估模型理论。然后，本文提出了适合我国的批量评估技术方案，依据我国各地CAMA系统实际建设与运行中存在的问题，提出并构建了综合特征价格模型和交易案例的自动评估模型，利用多元线性回归分析对模型进行校准。并以某市的存量房为研究对象对上述自动评估模型进行实证检验。通过一系列评估检验标准，本文验证了该批量评估技术方案适用我国的存量房市场，并且评估效果良好。接着，本文对神经网络中的一种计算模型——反向传播神经网络（BP神经网络）在房地产批量评估中的应用进行了深入的研究。根据房地产批量评估领域特征向量多、数据量大的特点，本文对反向传播算法固有的“收敛速度慢”的缺陷进行了改进，并对反向传播神经网络计算模型进行了实证检验，取得了良好的效果。最后，本文基于综合特征价格模型和交易案例的自动评估模型，采用J2EE/Java EE技术，实现了存量房批量评估系统，并详细介绍了系统中采用的关键技术。
[Abstract]:In the process of real estate value evaluation, there are many problems, such as different evaluation criteria, excessive cost and so on, which lead to the phenomenon of tax loss and social injustice to a certain extent. Many developed countries in Europe and America began to use computer-aided batch evaluation method many years ago. Computer-assisted Mass Appraisal evaluation (CAMA), which is based on the original cost method, market comparison method and income method evaluation principle, combines computer technology and statistical analysis technology to speed up the market value of a certain type of property in large quantities. A method of efficient evaluation. Now in developed countries in Europe and America, batch evaluation has gradually replaced single assessment and is widely used in tax base assessment of real estate. In contrast, batch evaluation technology in China is still in the initial stage of application. Thanks to the batch evaluation and promotion of stock housing started by the Ministry of Finance and the General Administration of Taxation in various regions in the second half of 2009, at present, CAMA systems have been initially established in most regions of our country. And on July 1, 2012, the transaction price of stock house declared by taxpayers has been comprehensively evaluated. However, the design of real estate tax base assessment system in various countries is closely related to the national economic structure, the establishment of government functional departments, the reform of tax system development, the legal system, and so on. The construction of CAMA system in China cannot copy the theory of foreign countries. In particular, the core of batch evaluation, automatic evaluation model (Automated Valuation Model), should consider the actual situation of our country, and localize and improve the mature batch evaluation technology abroad. In this paper, the concept and principle of batch evaluation technology are introduced in detail, the development situation at home and abroad is analyzed, and the concrete steps to implement batch evaluation are summarized systematically. The implementation of CAMA system has two very important key points, one is to build a comprehensive, true and accurate real estate information database, the other is to build an automatic evaluation model suitable for a specific property type. In this paper, the implementation of the above two key points is discussed, and the theory of automatic evaluation model is emphatically introduced. Then, this paper puts forward the batch evaluation technology scheme suitable for our country. According to the problems existing in the actual construction and operation of CAMA system in various parts of our country, this paper puts forward and constructs a comprehensive feature price model and an automatic evaluation model of transaction cases. Multiple linear regression analysis was used to calibrate the model. Taking the stock house of a certain city as the research object, this paper makes an empirical test on the above automatic evaluation model. Through a series of evaluation and inspection criteria, this paper verifies that the batch evaluation scheme is suitable for the stock housing market in China, and the evaluation effect is good. Then, the application of backpropagation neural network (BP neural network), a computational model of neural network, in real estate batch evaluation is studied in this paper. According to the characteristics of large amount of data and more feature vectors in the field of real estate batch evaluation, this paper improves the inherent "slow convergence speed" of back propagation algorithm, and makes an empirical test on the calculation model of back propagation neural network. Good results have been achieved. Finally, based on the integrated feature price model and the automatic evaluation model of transaction cases, this paper uses J2EE/Java EE technology to realize the stock house batch evaluation system, and introduces the key technologies in the system in detail.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP311.52;TP183

【参考文献】