Cluster Analysis Applied Research And Theory In The Determination Of The Port Size

Along with the development of the world economy and trade, modernization of port construction has been like a raging fire in China. There will be crucial strategic meanings to ascertaining port scale effectively and researching the methods to forecast port throughput and doing the prediction properly, in order to guide the programming and construction of the ports, to confirm the investment scale of the ports and moreover to accelerate the development of the areas. Traditional research on ascertaining port scale emphasize particularly on macroscopically programming and decision-making, which is affected by subjective factors; Traditional research on forecasting methods mainly focus on amelioration of methods themselves without choosing. This paper tries to use the method of demonstration research based on experiential data, analyzes a great deal of historical data, classifies ports of different scale into many levels through cluster analysis, studies the differences of increasing rules of the port's throughput ,and gets the conclusion of applicable forecasting methods of special port through the comparison and analysis.This paper expounds the fundamental of cluster analysis, studies the factors of port scale ,chooses index variables for port scale cluster analysis and 20 ports in our country to study .These ports are classified into 6 levels through the dissimilarity of port scale ,including international strong port, international big port, regional hinge port, regional big seaport .regional big freshwater port and locality replenishment port. These ports are also classified into 3 levels through the differences of increasing rules, including normally increasing port, accelerated increasing port and fluctuated increasing port. Finally, the author applies the time series models, grey models, regression model and RBF neural network model to forecast the port throughput of the three type ports, and gets the conclusion of applicable forecasting methods for special port through the comparison and analysis.