Advanced Predictive Model for Real-Time Fraud Detection of Cell Phone Purchase
A cloud-based solutions provider, acquired by the eighth largest retailer in the U.S., whose mission it is to unify the complex and interdependent data streams between stores, manufacturers, digital service providers and network operators.
Each year, the Telecom industry loses 15% of revenue due to fraud. Our client engaged with our team at Aditi to build a fraud detection module to determine the probability of a connected device or service purchase being fraudulent, operating in a similar fashion to the credit scoring system within the U.S.
Solution and Value Add
Our team of experts built machine learning models to detect fraud at real time. The prediction accuracy was between 80% and 90% between the three models, helping our client significantly reduce losses associated with fraudulent purchases and behaviors.