Evaluation And Prediction Of Coastal Zone Fishery Eco-economic System Based On Sgr And Rule Ensemble Method

Coastal zone is rich in resources, with the features of high productivity, great natural conditions, convenient geographic position and special property characteristic. Coastal zone has become the most economically invigorating region during human social and economic production activities. Coastal zone eco-economic system is very important. In China, coastal zone regions are the most rapidly growing economic regions. Because the economy develops fast and demand of resource is getting more and more, there are serious pressures on the coastal zone eco-economic system, such as large amount land-sourced pollutant emissions and unreasonable fishery. Therefore, it is necessary to do research on the relationship between the economic development and protection of environment.In this dissertation, coastal zone eco-economic system was analyzed by the Environmental Kuznets Curve Theory, non-parametric statistical methods such as stochastic gradient regression algorithm and ensemble learning methods such as RULE ENSEMBLE method. Main results and conclusions are shown as follows: First, according to Environmental Kuznets Curve (EKC) theory, measurement models of the relationship between economy development and environment quantity of the coastal zone city were established based on selected indicators of water quality, air quality and economic growth in coastal zone. Then the characters of the EKC were analyzed by the actual data of Tianjin. And policy recommendations were analyzed to ensure economic growth and improve coastal environmental quality eco-economic system provides policy recommendations.Second, the main factors of coastal fishery production were analyzed according to the nature-environment-social interaction. Using the pressure - state - response (P-S-R) model, a coastal eco-economic system index system was established. It is for the first time using stochastic gradient regression method to establish fisheries production model with the influence of stress factors, then the influence of each factor was analyzed. Using the actual data of Tianjin in recent years, there are analysis of the fishery output value of Tianjin and the major factors that influenced the degree. Finally, using the ensemble learning methods, the prediction model of fisheries production was established based on the ecological and economic system index, ARMA-GARCH model and RULE ENSEMBLE learning method. Then do some research on the development of Tianjin Fisheries production and the actual economic significance of the corresponding analysis. The results show that the RULE ENSEMBLE method performs well then support vector machine method and the adaptive regression spline method with better accuracy.