Research And Application Of Bank Customer Profile Construction Based On Big Data Technology

With the outbreak of COVID-19 and the emergence of many online new lifestyles and consumption patterns,user portrait technology has been widely researched and applied in recent years.And "Big data" has become an excellent carrier for enterprises to interpret users.Traditional commercial banks are also forced to conduct more detailed customer analysis under the background of digital transformation,the continuous upgrading of online financial needs and the challenges of Internet financial companies represented by Alipay.So this paper takes banking business as the research scenario and bank customer big data as the basis to study the construction and application of user portraits based on big data technology.And big data technology includes:big data preprocessing,big data analysis and mining,big data computing and other technologies.The main research work of this paper is as follows:(1)In order to solve the practical problems of banks’ current lending decisions being too much limited by human factors,time-consuming and labor-intensive credit evaluation and lack of scientificity,this paper constructs a"selection and identification" model based on the QSSVM algorithm for small and medium-sized bank customers.Customers are labeled as "good customers"who can lend and "bad customers" who cannot lend;This paper also designed a comparison experiment between QSSVM algorithm and QSVM,linear SVM,Gaussian kernel function SVM algorithm,and verified the superiority of the QSSVM model classification effect by using cross-validation on part of the ZS bank credit transaction data set.Finally,it also analyzes and explores the specific application of the label in the automatic credit extension of banks and the avoidance of credit risks.(2)For individual bank customers,in order to better understand customer groups,measure customer value and profitability,and meet diversified customer needs,this paper proposes a customer value segmentation model based on the improved K-means algorithm.According to the degree of activity and value contribution to the bank,the customer groups are divided into four categories:active important value customers,inactive important development customers,active important maintaining customers,and inactive general customers.In the selection of indicators for subdivision label construction,referring to the RFM model and combining the banking business situation,the RFMPA indicator system of the subdivision model is established;in the construction of the subdivision label model,the method of determining the initial cluster center by using outlier factors is improved.K-means algorithm;through comparative experiments,the sse index value is used as a measurement standard to verify that the improved clustering algorithm works well.Finally,the analysis discusses the formulation of targeted marketing strategies for these four different crowd labels.(3)In order to make the portrait label serve the business better,the customer portrait operating system is designed and implemented,including the label development module,the label storage module and the label application module.The system can realize tag creation,tag query and crowd selection,and has certain universality.