The Effect Analysis Of Commercial Bank Telemarketing Based On Optimized Xgboost Model

Nowadays,the process of globalization and liberalization of finance is gradually accelerating,the competition of banking industry is becoming more and more intense.It is difficult to continue to develop by the traditional profit model which relying on the interest spreads of deposit and loan.In the field of marketing,the trend is the traditional extensive customer marketing strategy transfers to the fine customer marketing strategy,and starts a customer-centric precision marketing activities.All bank marketing activities will rely on its huge data set.It is not possible to use manual work to analyze these data simply,and the data mining model is fully helpful for the analysis of these data sets.This paper aims at predicting the results of bank telemarketing.Firstly,it introduces the background and significance of the research,the current situation of research at home and abroad,and the research methods and ideas.Secondly,it introduces the data mining technology involved in this study,and the commonly used data mining algorithms such as Classification Regression Tree,Logical Regression,Random Forest and Gradient Boosting Decision Tree are introduced.On this basis,the xgboost integrated learning framework is introduced.And a borderline Synthetic Minority Oversampling Technique is introduced for dealing with unbalanced data sets(Borderline-SMOTE).The algorithm is combined with the above five kinds of data mining algorithms to establish the bank telemarketing classification model.Through the ROC curve,AUC value,sensitivity,specificity and other indicators,we found that Borderline-SMOTE algorithm combined with xgboost obtained the best predictive model with AUC value of 0.97.Secondly,xgboost model,whether in the prediction of the results or operational efficiency,are better than the other models constructed in this article.This paper also uses two kinds of information extraction methods(variable importance analysis and CART rule extraction)to extract the key information of the data set,and reveals several key attributes(for example,Euribor3 m,duration,age,etc.).This information extraction confirms that the obtained model is credible and valuable for telemarketing activity managers.