A data mining approach to the concession of e-commerce discount coupons: A case study

Abstract

This article aims to answer the following research question: ``How to build an efficient data mining solution for a discount coupon system?''. Thus, here a data mining solution is proposed to answer this question. The solution consists of four components: 1) use of the Random Forest technique as a classifier, 2) treatment of missing values, 3) enrichment of the database through the construction of new variables, and 4) use of the Kolmogorov Smirnov method to choose from the cut-off point for decision-making. An experimental study is conducted to validate the efficiency of the proposed solution. The results showed the acceptability of the Random Forest method to the problem and that the proposed knowledge acquisition strategy increased the predictive power. Finally, the results showed that the strategy of treating missing values has an influence on the discriminatory power of the solution. The contribution of this study is a guide to the construction of web-shop data mining solutions, giving guidelines on which data mining method to use, which is the best strategy for treatment of missing values, and how to improve predictive power through the acquisition of knowledge and how to choose the best cutting point.