A Research & An Applied Research On The Real Estate Appraisal Method Based On Self-Organizing Data Mining

Real estate appraisal in the overseas has more than 100 years history, but the development of our modern real estate appraisal only has more than 10 years. So far, research on the theory and methods of real estate appraisal in our country is at the exploration stage. Existing appraisal methods emphasize the experience and art, thus it has a very high request for the appraiser's personnel quality and correlated experience. With different subjectively judges and the different feeling to the same real estate, different appraisers' can work out a different appraisal result. On the other hand, research on the three traditional appraisal methods in the academic field is at theory research stage. To the appraisal operator, their research results and improvement seem difficult to understand and utilize. Therefore, how to consummate, improve the present general three real estate appraisal methods and establish scientific, operational, popular and quantitative appraisal method system has an important theory and practice significance.This article research contents mainly include following four aspects:First, on the base of analysis to the present situation and tendency of the domestic and foreign research on the real estate appraisal, through qualitative and quantitative analysis method, choose the main influence factors of real estate price.Second, establish the main influence factors system to real estate price based on the GMDH causality testing and built the good foundation for the real estate appraisal model.Third, in the foundation of traditional theories of the real estate appraisal, establish "the real estate appraisal model based on the AC pattern recognition method", and through the applied research and the comparison research, this new method is confirmed valid.Fourth, unify the model theory and actual appraisal work, produce "an operation sequence about real estate appraisal based on self-organizing data mining", and provide for the appraiser a concrete operational step.The research conclusions and innovating places have following 3 points:First, Introduced the self-organizing data mining method to the real estate appraisal for the first time, has established the real estate appraisal model based on the self-organizing data mining from the pattern recognition angle and produced an operation sequence for real estate appraisal based on self-organizing data mining, and through the applied research has confirmed it scientific and possibly operational, has enriched the real estate appraisal methods system.Second, comparing with three traditional appraisal methods, "real estate appraisal method based on self-organizing data mining" has 4 advantages: 1) the essence of based on the pattern recognition appraisal method is expansion of Market Comparison Approach. These election targets are more widespread, the samples are also richer, and this method can effectively reduce the subjectivity in the market comparison approach. 2) Comparing with Income Approach, it is not necessary to determinate the return to original estate interest rate and to forecast the future net-income of the real estate, the appraisal result is more objective and reasonable. 3) Comparing with Cost Approach, the fresh degree of the subject property will manifest through the room age target. Take the target of the room age as one of characteristic variables; the model can directly to obtain its shadow price and easily to obtain the accurate appraisal result. 4) Not only to a suit of the property appraisal, but also to a large amount real estate appraisal this method is more scientific, accurate and quick. To the large-scale appraisal of the government, for example, massive lands appraisal, this method also is operational. Third, comparing with the modern appraisal methods, "real estate appraisal method based on self-organizing data mining" has two big superiorities: 1) The request to sample data itself is not high. Fewer samples do not affect the scientific nature of the appraisal. 2) Melt the classical appraisal theory into AC model and enable it operational, higher credible and superior.