Machine Learning Use Cases in Fraud Prevention
simMachines supports use cases around fraud detection and prevention, charge back reduction, risk modeling and other related applications.
XAI Fraud Platform Automation
Problem
Large financial services firm uses a rule-based system for fraud detection globally and wanted to further automate and scale the process using machine learning.
Challenges
The client’s rule-based system provides explainability but requires manual maintenance that makes changes and updates problematic. Explainability is critical to managing fraud detection as part of their core platform.
Solution
simMachines XAI analytic workbench provides automation and continuous learning to update and manage fraud detection capabilities with less manual labor.
Result
Automation with precision was achieved, enabling the client to maintain explainability, improve results and reduce time and cost associated with managing fraud detection algorithms.
E-Commerce Fraud Prevention
Problem
Top 3 financial institution wanted to speed up ability to implement ML fraud detection solutions for e-commerce clients, enable continuous learning, and expose the factors driving the fraud.
Challenges
ML methods did not continuously learn or expose the Why factors. Increased competition was creating a need to upgrade speed and service value to client.
Solution
Ensemble approach using gradient boosting and similarity-based machine learning.
Result
70% increase over current fraud detection performance. Time to implement reduced from 6 weeks to 2 weeks. Continuous learning enabled and the Why factors reported/available behind every prediction.
Rewards Program Fraud Prevention
Problem
Large global retailer wanted to control its own fraud models for its rewards program, integrating with a 3rd party data platform.
Challenges
The data platform provider’s model building tools could not extend sufficient access to its client’s data science team.
Solution
simMachines branded its XAI Analytic Workbench as an extension to the partner’s data platform to enable client access for model building.
Result
After 1/2 day training, a single non-data scientist client resource created 8 models in 8 weeks that all matched or outperformed open source ML being tested internally. Transparency is a core value to the client and the XAI Analytics Workbench allows them to understand and explain their models’ behavior and enables investigation and detection of emerging fraud patterns.
Chargeback Prevention
Problem
Client needed to reduce charge backs in order to avoid triggering monthly order caps based on charge back volume.
Challenges
The current algorithms weren’t performing at the level required and lack of data science resources internally made it challenging to address effectively.
Solution
simMachines installed its XAI Analytic Workbench at the client and assisted in replacing current models with new charge back detection algorithms to stop predicted charge back order processing before it occurred, enabling more good orders to flow through.
Result
The XAI Analytic Workbench algorithm reduced the charge back rate enough to drive an immediate 2%+ increase in incremental order volume. simMachines trained an analyst at the client to use the software and develop new models for additional applications.
Digital Identity Resolution
Problem
Large scale DMP needed to improve linkages across devices associated with single users.
Challenges
Client had millions of unconnected devices it needed to accurately connect for improving campaign reach and frequency objectives.
Solution
The solution developed used similarity search to analyze a high volume of records quickly to pair two or more devices together with a high degree of accuracy quickly.
Result
The solution was able to comb through over a quintillion pairing combinations in 48 hours and pair over 800,000 out of 1 million device records together with ~95% accuracy compared to known ground truth. Daily updates can be run on a growing volume of records in under 2 hours.
Credit Risk Modeling
Problem
Large financial services firm uses a regression models for managing credit risk modeling and needs explainable machine learning to automate the process.
Challenges
Automation and performance lift with explainability made it difficult to find relevant machine learning alternatives.
Solution
simMachines XAI analytic workbench provides automation, explainability and achieves required performance levels against highly tuned regression models operating on an extensive data set.
Result
Automation with precision that matched required performance levels was achieved, in addition to being user friendly for modeling teams in build and deployment scenarios.
