| In the national economy,circulation industry,which is in series with multiple industries,occupies a strategic position.For one thing,it brings demand to production sectors and provides them with commodity outlets;for another,it supplies consumer sectors with goods and services.As the continuous change of economic pattern,the circulation industry is undergoing a profound transition.This thesis analyzes the feature of cycle fluctuation bases on climate index,and establishes early-warning model to estimate the direction and extent of the circulation industry’s development.First,this thesis comfirms the benchmark indicator of China’s circulation industry periodic fluctuation,that is the growth rate of circulation industry’s added value,and then,constructs the monitoring and early warning index system from three aspects:the overall development of the industry,self-support degree and macro support degree.Second,the benchmark index is decomposed into seven signals with different natural frequencies through EEMD decomposition method,thus we can identifies the constituent parts of circulation industry periodic fluctuation and their features.There’re two basic results listed as follows:(1)on the whole,the short-term shocks have a deep impact on the industry,according to it’s short term volatility,the development of this industry can be divided into three process whoes features are low scope-high frequency,high scope-low frequency and low scope-high frequency;(2)but in recent years,the adverse effects of mid-long term factors becomes the main reason causing the industry to fall.Third,after achieving circulant factor,cross correlation,peak-valley method and hierarchical clustering method for variables are used to mark out the the leading,consistent and lagging indicator groups.when clustering,chebyshev distance is chosen to measure the distance between variables and ward method is applied to calculate the distance between classes.For these three groups,three composite indices are drawn up;and comprehensive climate index is calculated use all indicators.Combing the trends of climate indices and the results of EEMD,we can obtain two results:(1)during the study period,circulation industry underwent five cycles,the first one is longest,the second one is of the greatest amplitude,and the last three are more smooth when China is transiting to a adjustment period after financial crisis;(2)in the next season,circulation industry will accelerate the decline base on the trend of leading index,and this fall is largely attributed to the glide of the growth rate of investment in the industry’s fixed assets and the recession of freight transportation industry.Last,comprehensive index is divided into five warning levels with ’μ-σ’ method.With comprehensive index as output variable,leading and consistent indicators as input variables,early-warning models are built base on BP-ANN,CART,CHAID and SVR which are suitable for continuous variables,mean while,boosting is applied into the first three models which are insufficient in small sample issue to rise accuracy.Then,the paper estimates the climate extent of the next season via polynomial kernel SVR,for it possesses great training and generalization ability.The result shows,the boom level will speed down and be on the edge of ’normal line’ according to the warning standard.Therefore,this thesis proposes several proposals:rationalize wealth distribution to boost consumption;take new technology into industrial process to meet market’s demands;innovate business model and merge the new market and traditional market to discover potential need;invest in transportation infrastructure and imformationize logistics warehouse. |
