Fraudulent Chargeback prevention probability analysis

🏷️  Name

Fraudulent Chargeback prevention probability analysis

🤵  Client

Ali K  (Turkey )

🏛️  Domain

Retail Platform

📅    Period

Apr-2024

📝  Description

View criteria of indicators leading to fraudulent chargebacks n E-Commerce by probability of  repeats of IP Address, Transactions failure counts, Email ID etc

📒  Achieved

This enabled my client in Greece to predict E Commerce customers who have high likelihood to be fraudulent to block them based on past record and quantity of a combination of factors. These could be usage of multiple IP addresses in an order ID, or Multiple EMail IDs in an order ID, or high number of transactions or a combination of 2 or more factors. Data was compiled and analyzed from a vast previous data set of over 70,000 transactions. This  enabled my client to form a set of rules to apply to his E commerce site. We gave a count and percentage probability of the past fraudulent transactions on these factors. The system had to be fine enough to give a high probability to block certain users without compromising to block genuine customers.  For example we gave data like customers who used more than IP address to log in or / and used more than 2 Email IDs in a failed E Commerce transaction had a 71% probability of being fake and could be blocked.