A Study On Influencing Factors Of Credit Card Consumption Installments Decision In China

China has become the second largest retail banking market in the world.Our commercial banks should seize the opportunity of consumption growth to boost the credit card business income as intermediary business.Our banks also need to expand the scale effect as foundations of transformation.However,the credit card market competition is also becoming increasingly fierce.At the same time,as an effective means to retain high-value customers and enhance brand influence and market competitiveness,precision marketing is increasingly valued by the banking industry.And with the continuous innovation and development of Internet technology,how to apply big data mining and analysis to explore new market opportunities and improve the effectiveness of prediction precision in the huge,diverse and complex credit card market is a worthy research topic.Based on the empirical context of consumer credit card installment,using the data of a commercial bank's 115612 credit card consumption and repayment from January to March 2018,this thesis compares and analyzes the data from traditional regression analysis and big data machine learning to forecast the consumer's credit card installment decision through logistic regression and support vector machine(SVM).At the same time,this thesis uses decision tree,random forest,and BP neural network as the robustness checks.The empirical results show that:(1)The traditional regression method has an accuracy rate of about 75% in predicting;and age,consumption amount,credit limit and financial situation are the factors to be considered in the prediction.(2)Compared with the traditional theory driven prediction method,machine learning can get a higher prediction accuracy(more than 90%).In addition,the PCA method can effectively optimize the algorithm.To sum up,based on the actual situation of China's consumer credit card installment decisionmaking under the background of big data,using appropriate machine learning algorithm can significantly improve the efficiency of predicting consumer decision making.