Prediction of Coupon Usage Behavior Based on Customer Segmentation and XGBoost algorithm

Coupons are an important means for merchants to promote sales. It is necessary to separately perform precision marketing on different types of customers and distribute coupons to customers who are more likely to consume. Therefore, a prediction model of coupon usage behavior based on customer segmentation was constructed in this paper. First, the sensitivity information customer makes on coupons was introduced into the RFM model to propose the improved RFM model, namely the RFS model. Moreover, combined with the K-means algorithm, the RFS model was divided into different categories according to customer value. Then, the XGBoost algorithm was applied to the full data and each category of customers to predict the write-off situation after receiving the coupon. Finally, actual cases were applied to predicting the use of coupons, which were compared with GBDT and LighGBM algorithms. The results showed that the prediction effect of the model proposed in this paper was higher than that of full data and other machine learning algorithms, which provided a reference for the accurate delivery of coupons in real applications.