Study On Regional Logistics Demand Forecasting Based On GA-SVM

As a very important part of regional logistics planning, regional logistics demand forecasting is the prerequisite and basis for comprehensive planning on the scale of regional logistics infrastructure, network spatial distribution, logistics business development direction and function positioning. Its aim is to provide logistics ability for regional logistics activities to meet continuous logistics demand, which can ensure relative balance between supply and demand of regional logistics services, so that regional logistics maintain a relatively high effectiveness and efficiency.On the basis of surveying current situation and existing problems at home and abroad, the paper firstly introduced concepts and features of logistics and regional logistics, and analyzed the concept, contents and steps of regional logistics demand forecasting; secondly, the paper identified main influencing factors of regional logistics demand, and, built an index system of regional logistics demand forecasting from regional economic output, regional industrial structure, regional consumption level and so on; thirdly, genetic algorithms (GA) was introduced into support vector machine (SVM) model to optimize its kernel functions and parameters, which produced an improved SVM model based on GA (GA-SVM) and the algorithm was carried out in MATLAB; Finally, the paper used GA-SVM model to forecast logistics demands of Handan City, and compared forecasting results with those by traditional support vector machine and BP neural network to prove the feasibility and effectiveness of the improved algorithm, providing a new method for regional logistics demand forecasting.