Metropolitan Logistics Demand Forecasting Based On Support Vector Machine Optimized

With the advent of the era of big data, it aroused many scholars’ machine learning enthusiasm. Logistics Demand Forecast in the past has always been a hot and difficult in the social research. In the past, because of insufficient data, scholars can only carry out the logistics needs of a wide range of areas of study, while the study of the demand for Logistics of the metropolitan area is not a lot. For the local governments, the scale of logistics in the region let it know their logistics status quo jurisdictions. For metropolitan economic group, the collected data form the near last years is often not perfect, representative data is not strong, while they often more multi-dimensional record of the local government, the breakdown of the scale of logistics demand will be a certain influence. Thus established, operational logistics demand indicators and the corresponding essential scientific prediction model based on regional characteristics is very important Discussing the metropolitan forecasting of logistics demand issues is from the following parts.The first part, focusing on the status of current research issues at home and abroad Metropolitan logistics needs to sort out and analysis; we found the problem and ask questions. Then it describes the purpose and significance, main content, articles frameworks, research methods and innovative points.The second part, the main metropolitan area of logistics and logistics requirements are summarized, analyzed the functional elements metropolitan of logistics. On this basis, the metropolitan of logistics demand characteristics were analyzed, and a detailed analysis of the factors affect metropolitan logistics demand.The third part, for the characteristics of metropolitan logistics demand, primarily by the process of the current literature Metropolitan Logistics Demand indicators were analyzed with a number of quantitative indicators finally elected representative; and in the selection of the relevant under the principle, combined with the actual situation collate, analyze and put forward three latitude indicators:economic scale index scale logistics capability index, population size indicators. Three indicators include a total of nine quantitative indicators more representative.The fourth part, focusing on the current prediction methods at home and abroad were analyzed and compared, it chose the most common prediction, neural network prediction, support vector machine analyzed theoretically. Finally factors affecting metropolitan logistics demand and characteristics of selected quantitative indicators SVR most predictable way.The fifth part is mainly based on the characteristics of the current status of research and Metropolitan logistics demand indicators, the first principal component analysis of a large number of quantitative indicators were lower treatment, making training more data fit to support vector regression prediction model, which reduce the training time and avoid repetitive; on this basis, the paper proposes the use of inverse hyperbolic sine function SVR optimized, thus improving the accuracy of prediction and optimization of the parameters to shorten the time; finally, through the logistics needs of the Chongqing metropolitan empirical analysis to test the accuracy and feasibility of the model.