The Research Of Theory And Demonstration In Synchronous Optimization For Personal Credit Rating

Credit is one of the important problem that concerns healthy development of the market economy, and is also the main obstacle that influences consumption of our country. With the development of the consumptive credit loan, commercial banks have to focus on how to avoid personal credit risk and improve efficiency for more profit. Therefore, A suit of precise and practicality personal credit rating method is needed for adapting the market competition of loan business of commercial banks.The paper joints the theory and the practice. On the basis of abundance of useful information about personal credit rating, methods of personal credit rating are summarized by concretely analysis. Taking the characters of personal credit rating into account and aiming at the shortage of history data sample in our country at present, statistical learning theory and support vector machine are used for establishing the intelligent model. In order to combine theories and fulfillment, we used actual data to do experiments and showed better result.The content of the paper is organized as follows:First of all, the paper introduces study background and meaning. Then, domestic and international condition is summaried at present. Secondly, the paper includes the concept and the theories of personal credit rating. We elaborate the concept and method of personal credit rating. Then, process and principle is summarized. Methods of personal credit rating related to the texts are discussed and compared. Thirdly, statistical learning theory, support vector machine and genetic algorithms are introduced. We advanced the method that can be used to optimize index system of credit assessing and parameter of model at the same time. Through utilizing the reliant relation of index system and parameter, the system of credit assessing made progress. Consulting some production, we carefully design the genetic algorithms and establish the rating system. Fourthly, through the experiment in different types of support vector machine, we can observe the influence of result and analyse the mutative process of parameter. We make sure that genetic algorithms is useful during the process looking for the superior solution. Finally, the paper includes summarization and conclusion. We make a summary and put forward a series of viable suggestion. We bring forward the direction of future research.We do experiment and find that the result of using C-svm and Nu-dvm is equal.