Macro-Economic Early-Warning Wethod Reaseach-Kohonen-Bp Neural Network Model
|School||Zhejiang Technology and Business University|
|Keywords||Macroeconomic early warning method Methods are summarized research Kohonen neural network clustering BP neural network model|
In the review based on domestic and international macroeconomic warning method , various methods summarized as the traditional index early warning method , four major categories of the traditional boom semaphores law , warning method of measurement model early warning method and pattern recognition , from first principles , excellent the inferiority , suitability angle made ??of four types of methods the method theoretical studies suggest possible improvement ideas . The study further deepening macroeconomic early warning method to provide a reference basis for an accurate understanding of scientific and rational use of various methods , and also pave the way for theoretical model innovation made ??. This article keep up with the neural network in the macroeconomic the warning field use and development of the new hot spots , in-depth analysis of the topology of the Kohonen and BP neural network model , algorithm theory , model characteristics and application areas , made ??a comparison of the two models are different at the analysis , based on a warning model based the Kohonen-BP neural network , and build a model of the structure of processes . The basic idea of the model : using Kohonen cluster analysis of the data , and the results obtained with experts warning degree given category analysis and correction , so the corrected sample data input BP neural network training , and finally use The neural network output warning analysis . Its main purpose is to further deepen the objective of macroeconomic warning , strengthen its handling complex and highly nonlinear economic warning capabilities . Finally, build the macroeconomic early warning indicator system , using the model made ??empirical research . The results show that the model has a good prediction . Full process using the SPSS Clementine12.0 operation .