Promo Abuse Modeling in E-Commerce Using Machine Learning Approach
Tuga Mauritsius, Sofia Alatas, Faisal Binsar, Riyanto Jayadi, Nilo Legowo · 2020
To increase the attractiveness of new customers and customer engagement, e-store makes various promos. However, as faced by XYZ, an e-commerce company based in Indonesia, some of the promos are prone to be misused by some users. A well-known type of abuse is the reduplication of accounts by the same user to get more coupons or promotions fraudulently. This abusive act can bring huge losses to the company, as the promotional campaign will miss the intended targets. The purpose of this study is to examine whether some data mining techniques such as the J48 Algorithm and Random Forest Algorithm can be deployed to detect promo misuse based on available customer's profile. The study uses a dataset from the transaction history recorded by XYZ company during the years 2018 and 2019. The results show that both algorithms can accurately model the FRAUD on the promos where the Random Forest Algorithm obtain a significantly higher level of accuracy.