Research On Precision Marketing Of Bank Credit Cards Based On Data Mining Technology

As the credit card industry shifts from the "era of increment" to the "era of stock",the contradiction between the diversified and individual needs of customer groups and the homogeneous and standardized crude marketing methods of banks has become increasingly prominent.Coupled with the huge impact of Internet finance on the traditional bank credit card industry,the bank credit card industry is facing the dilemma of inefficient marketing and declining efficiency.As a result,banks have started to deeply integrate with big data at a low growth rate.How to rely on big data technology to achieve a win-win situation for both customers and banks and build a new profit growth engine has become one of the primary tasks of banks at this stage.Based on data mining technology,this paper focuses on how to use deep learning network to deeply analyse and accurately mine the bank's existing massive customer data,so as to identify quality customers that meet the bank's value needs,and carry out precision marketing models based on the predicted results of high-quality customers.Ultimately,a precision marketing system of acquiring customers-live customers-retaining customers is created to dynamically monitor and predict customer behaviour at each stage of the customer life cycle,in order to optimize the bank's strategy and provide personalized services to customers.Specifically,the main research elements of this paper are as follows.(1)The overall design of credit card precision marketing system based on different characteristic attributes.To address the problem of constructing the portrait label of bank credit card quality customers,the overall requirements of banks for credit card customers and customer relationship management methods are studied,and the dual perspective of customer creditworthiness and customer loyalty is proposed to reflect the degree of customer quality.Considering the subjective bias brought by relying on manual annotation on the classification results,this paper adopts the unique tangible behavioral data as the statistical label and the five tangible underlying data of personal attributes,asset attributes,risk attributes,consumption attributes and product attributes as the driving factors to construct a labeling system for quality credit card customers.To address the feature selection problem,the Light GBM algorithm is used to calculate the importance of the underlying features,which is used as the basis for validity feature selection.Regarding the data quality problem in the bank database,this paper proposes a targeted data governance solution to provide a strong guarantee for model accuracy.(2)Research on the prediction model of potential high-quality credit card customers of banks.Considering the sparsity of data,the complexity of manual feature derivation and the interpretability of feature influence,this paper introduces the Factorization Machine,Attentional Factorization Machine,A Factorization Machine based Neural Network,which have achieved good results in the field of CTR to segment existing bank customers,unearth potential quality customers and identify those who need to be maintained in order to achieve accurate customer acquisition.In order to solve the problems of balance of missing information and run rate,and to reduce the noise generated by invalid interaction features,Deep AFM is proposed to improve the algorithm.The four models are finally experimented and compared to validate their applicability and effectiveness on the bank's quality credit card customer prediction model.(3)Credit card customer precision marketing interactive system design.In order to achieve data-driven marketing,and use the results of deep mining of high-quality credit card customers to assist precision marketing decisions to solve practical business problems,this paper builds a precision marketing interactive system based on Python and My SQL to provide marketers with accurate customer segmentation results and dynamic monitoring of customer information to achieve accurate and personalized marketing decisions,while adapting to changes in demand in a dynamic market.