Research On The Targeting Of E-commerce Coupon Based On Bootstrap Methods

With the fierce development of international e-commerce, comes the big bang of all sorts of users information, and the big data era has come. How to utilize this enormous data to serve internet marketing, this topic has attracted lots of attention. This article is about the accurate marketing with the big data. This case is about the coupon sending out, how to target the user group based on their profile to maximize the redemption, therefore boost its ROI. Firstly a Logit model was estimated by the samples, and selected the strongly relevant ones from 18 variables, then calculated ROC and AUC to evaluate the prediction ability of this model. Then by utilizing bootstrap method, estimate the variance of AUC and its confidence level to validate this model’s classification ability. In addition, also find the best threshold of classification probability to maximize the Sensitivity and Specificity, besides, their confidence levels were also estimated with bootstrap methodology.