This repo contains data and R code for Order Brushing Detection Task in E-Commerce Industry, part of the 2020 Shopee Code League Competition hosted on Kaggle.
-
If you not familiar with the term, it refers to seller misconduct activity done by creating fake transactions (i.e buyer-bot). This act, according to Shopee's Terms of Service, is illegal.
-
The main indicator is transaction volume.
-
If the transaction volume done by seller-buyer reached a certain threshold (for our purposes, let's say it's 3) within a one-hour interval, then we can suspect the seller might have done order brushing.
-
If that's the case, then we put seller ID and list of buyer-bot (with format "buyer1&buyer2&buyer3&...&buyerN") side by side. Else, we put seller ID and "0" side by side.
This task is important because we want to protect the integrity of the endorsement system within said E-Commerce company's website.
Pardon my messy writing on several lines of code.
Upload short tutorials in the future when time permits.