A Forecasting System For Aquatic Products Price Based On Non-parameter Model

A Forecasting System for Aquatic products price is a significant measure to perfect and establish the aquatic products wholesale market and evade risk. It also can be helpful to guide aquatic production, facilitate aquatic circulation and help the production and management achieve a well economic and social effect. So we put forward to study on a forecasting system for aquatic products price, which can provide forecasting information service for aquatic products marketers. We have fulfilled five aspects of the research.1) The theory "The weak form of efficient market" has been validated not tenable for aquatic products wholesale market by using the method YOUCHENG checkout and R/S analysis by wholesale price of aquatic products between 1997 to 2002 from BeiJing XinFaDi market, The results shows that there is some relativity among price income, price can be forecasted by price time series.2) Wholesale price fluctuation rule of aquatic products has been researched between 1997 to 2002 from Beijing XinFaDi market. From power spectrum analysis, P(f) ∝ f-β indicates that wholesale price fluctuation of aquatic products presents single fractal character. The planar relation graph of Dq and q is not a beeline by GP arithmetic, which shows that wholesale price fluctuation of aquatic products presents multi-fractal character. This discovery tells us that multi-fractal character is an integrated cause of wholesale price fluctuation of aquatic products.3) How to obtain multi-fractal character of wholesale price fluctuation of aquatic products has been researched, and results showed that it is contained in wavelet transformation. Mulit-fractalcharacter has been obtained by sample h3 (1/16, 1/4 , 3/8 ,1/4 , 1/16) and A Trous arithmetic.4) Lyapunov exponent and Renyi entropy of wholesale price time series of chaos character of time series has been analyzed, which proved that price time series is a chaos time series. The max and average time scales for forecasting have been attained from Lyapunov exponent and Renyi entropy.5) A hybrid model has been built integrated wavelet neural network with genetic algorithm.6) A Forecasting System has been designed and achieved.