Research Of Forecast Of Credit Card Customer Default Based On Improving Intelligent Algorithm

In 1960 s,intelligent algorithms became popular which are created by someone are inspired by the laws of the natural world and imitated the laws of the nature.Intelligent algorithms differ from gradient-based algorithms in that they rely on probability to search.The successful application of intelligent algorithms in solving practical problems lays a steady foundation for its development.Since the birth of credit cards,there has been several decades of development history.The amount of credit card holders has increased year by year in the world.Nowadays,credit cards are playing a very important role in people's lives.People are focusing more and more attention on the development of credit cards.Credit cards provide people with quality services while also causing some problems.one of problems is how to assess whether a customer is breaking a agreement which is troubling people.This article analyzed and researched the credit card customer data by intelligent algorithms and data analysis technologies,expecting to find out the factors that have a greater impact on the credit card customer's default and establish an effective classification model.this paper predominately completes the following four areas:1.The study on the issuance use of credit cards,the types and function of credit cards at home and abroad,also analyzing the future development of credit cards.2.The basic theories of intelligent algorithms are studied including particle swarm optimization,gray wolf optimization,genetic algorithm and BP(back propagation)neural network algorithm.This research introduced fuzzy set theory,mean impact value method and support vector machine,which established the MIVBP-SVM model and the IFBPNN model with intelligent algorithms.3.The data processing technologies are used to process the credit card customer,the attributes of this data set are analyzed respectively including age,level of education,marital status and overdraw.The data shows that among the cardholders,the single person is higher than the married person.In terms of educational level,university students account for 47.30% of the total and graduate students account for 35.69%.From the respect of breach of customer data,singles accounted for 54.14% of the total breach of contract.4.In this paper,MIVBP-SVM model,IBPNN model,BP model,SVM model,GABP model,and PSOBP model are used to predict credit card customer default,and the results are analyzed and discussed,which reflect that the IFBPNN model has a best effect.