Design For Customer Risk Evaluation System Based On Multi-class SVM

With developing of market economy and international integrating, the management environment is changing day by day. This kind of change turns more complex and more indefinite, takes more hidden dangers and more latent risks to the enterprise's management.So it's necessary to enhance research on marketing risk, strengthen marketing safe consciousness, establish a scientific marketing risk warning system, and form a complete enterprise marketing risk guard mechanism.In the marketing risk recognition method, this paper introduces a new statistical study model. Support Vector Machine (SVM), based on the counts study theory, is a research hot spot in machine learning domain. It has good performance on regarding two classified questions with small samples, and also has the overall situation optimality and good exudes ability.This paper focuses on the marketing risk evaluation, and introduce the method of Multi-Class Support Vector Machine as a new appraisal model to the marketing risk assessment practice.By using the object-oriented design method, this paper implemented modulation seal of Support Vector Machine algorithm, achieved the goal of marketing risk evaluation with Multi-Class Support Vector Machine in database software.In addition, in order to make the software system have better maintainability and more nimble function development, this paper constructs a plug-in management system, and establish the software's function module by a plug-in unit form.Thus, we achieved the goal of stripping the function from the software master routine, reduced the software maintenance complexity, and enhanced the software flexibility to a great extent.