Risk Assessment And Application Of Real Estate Investment Based On GABP Neural-Network And SVM

Real estate development is a high investment, high economic return, high risks investment activities, subject to the social, economic, technological and other factors influenced. There is uncertainty factors in the development process, for investors, with high returns possible, also contains a corresponding high risk. And many investors see only the real estate market to bring high return on investment, but they ignore the risk after the high economic return, analysis of project risk with subjective experience during the investment decision-making, very few systematic risk comprehensive analysis, evaluation, the event could easily result in investment failure when risk factors happened. Therefore, to study the real estate development and investment risk assessment methods has important theoretical and practical significance.In this paper the main achievements are as follows:first, explained the relevant theories of the real estate development and investment risk analysis, about the contents of real estate investment risk analysis; Secondly, describes relevant theory of the BP network, genetic algorithms and support vector machine; This paper focuses on using examples of analysis of GABP and the SVM method was applied to the effect of prediction of the real estate risk, because of our predecessors have done research GABP in many areas, has also been some research on the real estate, therefore this also introduction of a new method—support vector machine (SVM) and its application to real estate investment risk evaluation. This method is relatively new, preliminary study in a number of areas, but basically no research in the real estate, this paper attempts to established the model of the real estate investment risk assessment based on SVM, comparative study of two methods, focusing on the extraction method of SVM prediction effect, empirical results show that although both methods can be applied to the field of real estate risk prediction, but the SVM is superior to GABP, and a higher prediction accuracy can be applied to risk assessment in real estate.In this paper, according to the shortcomings of BP network, established the model of the real estate investment risk assessment based on GABP and SVM, GABP is to ptimize learning the neural network weights and thresholds use genetic algorithms (GA), re-use of the optimized BP network training samples again, and finally get the optimal solutionof the problem. Support Vector Machine is a new type of machine learning algorithm, which can be very successful in dealing with classification and regression problems. Its good non-linear quality, high fitting precision, flexible and effective way of learning, does not depend on the characteristics of the sample, made risk prediction of real estate investment is good.The dissertation provides a practicable method for the real estate investment risk prediction, which is important for our real estate circle to improve the level of investment decision-making.