Font Size: a A A

Research On Mass Appraisal Model Of Real Estate Based On Support Vector Regression

Posted on:2014-10-14Degree:MasterType:Thesis
Country:ChinaCandidate:Z Q GuoFull Text:PDF
GTID:2309330392963949Subject:Management Science and Engineering
Abstract/Summary:PDF Full Text Request
Due to the development of the real estate appraisal industry is not mature enough,the quality of the relevant employees, the valuation methods, and the factors thataffect the Real Estate prices are not fixed factors which may contribute to the widedisparities in the appraisal. A scientific solution is needed for the standardizedmanagement of valuation industry. Based on the Hedonic valuation model, the studyworks out a reasonable and perfect mass appraisal model of real estate using GIS andsupport vector regression. First of all, the housing data is collected according to thedifferent real estate and the different type of housing Transactions, as well as thelocation factors surrounding the real estate is extracted through GIS system called bythe Google Maps API. Then, Rates for different types of housing are amended to theStandard Rates under the same physical condition, and the real estates are clustered bythe different price of them, which are divided into the low-price housing, theaffordable housing and the high-price housingļ¼Œand the sample is split into trainingand test sets for each type of real estate. Afterwards, for housing training set of eachtype, respectively, with four kinds of kernel functions, and the four categories ofparameters extracting, such as the empirical parameter method, the cross-validationmethod, the genetic algorithm, and the PSO algorithm optimization method based onadaptive mutation. The support vector machine regression method established themost accurate valuation results using radial basis function and the PSO algorithmbased on adaptive mutation, and the corresponding prices are predicted through thebulk valuations on the known location factors in test sets, but also to test the accuracyof the support vector machine regression model. The valuation results aresignificantly better than the ridge regression and BP neural network. Finally, thebulk-valuation model is comprehensively evaluated through the evaluation indexsystem provided by the International Assessor Association (IAAO). And the massappraisal of real estate has developed a standardized workflow.
Keywords/Search Tags:GIS, Support Vector Regression, Mass Appraisal Model of RealEstate, PSO Algorithm Based on Adaptive Mutation
PDF Full Text Request
Related items