Fraud Detection System
Transactions are sent to Vesta API and real-time decisions are made by machine learning models using data from behavioral analytics, merchants, third-party vendors and advanced feature engineering.
FRAUD ANALYSIS DATA WAREHOUSE
Transaction details, supporting data elements, decisions, chargeback information as well as additional data that were not available real-time are further collected and securely stored in our data warehouse, allowing our data scientists and fraud analyst to see potential fraud from every possible angle—continuously improving our modeling and decisioning.
Machine Learning Pipeline
Models with cutting-edge machine learning algorithms are trained automatically and constantly updated in our decisioning platform for optimal performance.
Fraud Analyst Review
Our fraud analyst team strategically selects a sample of transactions to ensure that the models are working as trained. Additionally, they also analyze anomaly alerts from our unsupervised learning models to continuously improve the performance for our merchants and partners.