Study On The Real Estate Appraisal Based On The Support Vector Machine

In recent years, along with the real estate primary market and secondary market's active development, and real estate transactions have become increasingly frequent, so as determined prices of real estate transactions in the real estate transaction valuation are attracting increasing attention. Because the influence factors of the real estate price is complex and inconstant, they must be considered both the characteristic of the real estate market and the influence factors in the appraisal of the real estate. So apart from the corresponding basis for the theory and methods, it must also rely on the personnel changes in the market valuation of the ability to grasp the extent and judgment.Market comparison approach, cost approach and income approach are the mostly method of the appraisal of the real estate, which value the real estate from the different angle, it has subjectivity and limitation. The machine learning via the learning of the data, search the function relation of the input and the output, and using the function forecasts the output for the random input. The support vector machine is a new machine learning method based on the statistical learning theory. It embodies the very important principle in the statistical learning theory, which is structural risk minimization (SRM). It solves the fitting of the nonlinearity, limit data, and have the better generalize and the steady frame. So SVM is applying in the different areas.The article put forward to applying SVM for the appraisal of the real estate on the base of analyzing three basic method of the appraisal of the real estate. It introduces the concept, theory and method of SVM, and setting the criterion of the scalar. And it repeats trying to calculate for the Xi'an 2001-2006 year actual transaction's 60 groups data, finally setting the model of the value of the real estate. The article forecasts the 10 groups, and gains the result which has the better precision. For testing the availability of the model, it analyses the forecast result of the market comparison approach and RBF, finally sums up some meaning conclusion.