Research On Customer Profile Model Of A Commercial Bank Based On Machine Learning

Customers are essential to commercial banks.Facing with fierce industry competition and saturated market environment,how to effectively use customer data analysis to target marketing customers is very important to improve bank income.Machine learning is an important meaning to obtain valuable information from data.Using machine learning technology to model the customer data of commercial banks can classify customers more accurately.This paper analyzes the customer data of commercial banks based on machine learning methods,builds a commercial bank customers profile model,and applies the customers profile model to the bank customers profile system,so that the profile system can display customer information more comprehensively Banking staff analyzes customers more easily.The main work of this article includes:1.First,pre-processing operations are required on the initial commercial bank customer data,and the generative adversarial network(GAN)is applied to bank customer data processing to deal with existing the category imbalance of bank customer data,because the generative adversarial networks can better simulate the original sample distribution and generate high-quality samples than traditional methods.At the same time,the decision tree is used as a classifier,and the GAN model proposed in this paper is compared with the commonly used unbalanced processing methods such as random undersampling(RUS),random oversampling(ROS),SMOTE,BSMOTE and ADASYN.The results show that the generative adversarial network model proposed in this paper has a better effect on the comprehensive measurement of indicators such as accuracy rate,recall rate and F-measure for category imbalance problems.2.Construct a customer profile model based on the processed customer data.When building a customer profile model,this paper uses a combination of two data models to solve the problem of high-dimensional bank data that is difficult to classify customers.First,use more explanatory factor analysis to reduce the dimension of the pre-processed data.Then use the FCM algorithm to build a customer segmentation model,cluster out six categories of customer results,and use the CS algorithm to optimize the FCM to improve the model's convergence speed,learn the required data features quickly and accurately,and finally based on the factor weights to summarize and describe each category.Next,constructing an adversarial network based on the processed data,the customer churn classification is carried out to obtain a customer churn warning model.Taking indicators such as F-measure as the measurement standard,the best warning model is selected from three methods:Random Forest,Xgboost,and LightGBM.Finally,two models of the above are combined to obtain the customer profile model in this paper,so that customers can be accurately divided into multiple categories.3.In order to facilitate the direct acquisition of bank business personnel's analysis results of the bank's customers,based on the analysis results of the constructed profile model,design and implement a commercial bank customer profile system,which can automatically analyze commercial bank data,provide query,warning and visual display of profile.And this system can satisfy the business staff's analysis of the customer's data,which is beneficial to improve the bank's efficiency.This paper uses a variety of machine learning methods to study the commercial bank customer porfile model,so that high-dimensional and complex bank customer data can be more accurately classified.Among them,the GAN model is applied to class imbalance processing,and through optimization of the fuzzy mean algorithm to improve the overall effect of the profile model in this article.Finally,the division of customers in the profile model of this paper is applied to the bank profile system,which realizes the automation and visual display of data analysis.