Research On The Influence Factors Of O2O Coupon Use Based On Bayesian Network

Consumers' behavior habits have also changed drastically,and various marketing are competing,among which the O2 O business,which organically combines the Internet and offline business channels,is developing most rapidly.At present,there are abundant researches on the technical support and actual practice of O2 O business,while there are few researches on the factors of the utilization rate of O2 O coupons.Analysis of O2 O coupon usage factors can not only help merchants develop more efficient promotion strategies,but also help consumers shield interference from various promotional and preferential activities,accurately judge the actual discount value of coupons,and obtain real discounts.In this paper,real O2O consumption record data provided by an e-commerce platform are used to split the data based on the sliding window method.On the basis of the original record features,feature engineering mining data information is comprehensively constructed from multiple angles to extract 88 new features.On this basis,the XGBoost was trained,and the training AUC is 0.9008,and the verification AUC is0.8887.The AUC is submitted on the platform,and the average online AUC is 0.8003.In order to build a more intuitive and efficient Bayesian network,dimensionality of the above information is reduced according to different modules,and eight comprehensive features of O2 O coupon use are obtained: Users' consumption activity degree,users' preference of coupon intensity,users' consumption convenience requirements,merchants' business volume,merchants' own user penetration,merchants' preference intensity habit,coupon ease of use and users' historical preference degree in specific merchants.Based on the analysis of constructed Bayesian network and XGBoost,followed are the conclusions: 1.The three factors that most directly affect the use of O2 O coupons are the business volume,the habit of discount strength of merchants and the preference of coupon strength of users.O2 O merchants should focus on these three aspects when developing coupon marketing strategies.2.The user's activity associated with the use of coupons,in bayesian network structure,use coupons for reason,user is active for the result,this suggests that the user is not active more easier for the use of coupons,coupons instead of focus on the lead to more active users,so should reduce the number of active users when put in the coupon on the weights.In this paper,feature engineering is carried out from multiple perspectives,and the machine learning is used to verify the information adequacy,and bayesian network is applied to study the influencing factors of O2 O coupon use for the first time,which provides theoretical basis for O2 O merchants to formulate promotion plans efficiently and improve coupon marketing revenue.