Combination Forecasting Model Based On Support Vector Machine And Application In Personal Credit Evaluation

In recent years, with the sustaining development of domestic economy and the rapidchange on people’s expense perception in China, each kind of personal credit consumptionalso continuous growing fast. Commercial banks start to take consumer loan which iscontinuous heated as an important component of their future development strategy. However,These banks and some credit information services don’t have the reasonable standard,evaluation method and related system on personal credit consumption management. In theprocess of dealing with business, they still use empirical methods to evaluate the personalcredit consume applicants’ credit condition and ability to repay money on time. At present,in the academic circles, traditional statistical analysis method and machine learning methodare used in personal credit evaluation, but they can’t find a perfect equilibrium pointbetween accuracy and stability.Based on the research result of domestic and foreign about the personal credit, thispaper builds an acceptable model for intestine personal credit evaluation by supper vectormachine which is widely used recently, and use credit loans data of a commercial bank toexamine the model. First of all, on the condition of reviewing the pertinent literature athome and abroad, the article constructs an acceptable for index system of our country. Then,a supper vector machine model of combined kernel function is built to avoid thedisadvantage of single kernel function. In order to avert the divergence caused by wrongassortment of different kinds which would cause badly effect on classification model, herewill introduce a combined cost-sensitive support vector machine. Later, the traditionalstatistical analysis model of supper vector machine and neutral network is in cooperation tostructure a better system than single model and the conclusion is that the model is morepractical.