Abstract: Real estate is a business of high risk. This paper establishes an optimized MIV-BP neural network(Mean Impact Value Back-Propagation Network)which is based on a successful Back-Propagation neural network to identify the risk of real estate projects and to analyze the influence of various factors in the risk of real estate projects, thus to provide some references about the risk recognition for the real estate projects investment decisions and to help the real estate companies to avoid the risk effectively. Some present data related real estate projects are adopted to test the accuracy and objectivity of this model. The test results show the MIV-BP neural network model has an excellent compatibility and more accuracy when it is used in the risk recognition of real estate projects which can meet the experts' evaluation requirements and has a good application value in the analysis of risk factors in real estate projects.

Key words: real estate, risk recognition, MIV-BP Neural Network, MIV Algorithm

摘要： 为了更准确更客观地识别房地产项目中的风险,为房地产项目投资决策提供科学依据和参考,有效地规避风险,本研究在BP神经网络 (Back-Propagation Neural Network)建模的基础上,采取MIV(Mean Impact Value)算法对BP神经网络模型进行变量筛选的网络优化和改良,从而形成新的优化后的MIV-BP(Mean Impact Value Back-Propagation Neural Network)神经网络,并以此用于评价房地产项目中的风险度以及各因素在风险度中的影响作用大小;同时选取目前相关的房地产项目数据进行仿真实证分析和验证。验证实验结果表明,MIV-BP型神经网络对于房地产项目风险度识别具有良好的适应性和准确性,实验结果客观,达到专家评价的要求,并在风险因素作用度分析上具有良好的应用价值。

关键词: 房地产, 风险识别, MIV-BP网络, MIV算法