Understand what Amazon Fraud Detector is, how it works, and how to use it for your business use case.
Amazon Fraud Detector Documentation
Amazon Fraud Detector is a fully managed service that makes it easy to identify potentially fraudulent online activities such as online payment fraud and creation of fake accounts.
In a few steps, you can create machine learning models to identify a variety of fraudulent activities.
Getting started
- Before you get started with Amazon Fraud Detector, set up the interfaces you plan to use.
- Use the tutorial to get started with Amazon Fraud Detector using the AWS Management Console or the AWS SDK for Python.
Build and train fraud detection model
- Choose a model type for your business use case and gain insights into the dataset requirements for the model type.
- Gather data from your business for your use case with the help of the insights provided for the model type you have chosen.
- Define the event that you want to use for evaluating fraud. An event is the business activity that you want to evaluate for fraud risk. An event type defines the structure of the individual event.
- Store the data that you gathered from your business for the event that you want to evaluate for fraud. You can store your dataset internally with Amazon Fraud Detector or externally with Amazon S3.
- Amazon Fraud Detector builds a fraud detection model using the model type that you choose for the event type that you define. The model is trained using the dataset that you prepared.
- As part of the model training, Amazon Fraud Detector generates a model performance score and other model performance metrics. Use the metrics to evaluate your model performance.
Fraud detection, prediction, and downstream processing
- Create a detector by adding your trained model and rules for detecting fraud. Detector is a container that contains a fraud detection model and rules, for one specific business event you want to evaluate for fraud.
- Deploy detector in your production environment.
- Get fraud predictions for a single event in real time or get fraud predictions offline for a set of events and then use prediction explanations to gain insight into how each variable that you use in your dataset impacts your model's fraud prediction score.
- Set up orchestration to send your events data to other AWS services for downstream processing of the events after fraud prediction.
Model resources, security, and monitoring
- Learn how to manage the resources that the models and detectors use to evaluate your business events for fraud. Resources can be variables, rules, outcomes, labels, lists, and entities.
- Understand how to configure Amazon Fraud Detector to control access to resources, to protect your data, and to meet your security and compliance objectives.
- Use monitoring tools that AWS provides to monitor Amazon Fraud Detector, report when something is wrong, and take action when appropriate.
References
- Describes all of the API operations for Amazon Fraud Detector, with sample requests, responses, and errors for the supported web service protocols.
- Describes all of the classes that are included in the Amazon Fraud Detector AWS SDK for Python (Boto3).
- Describes all API operations for the AWS SDK for JavaScript, with sample requests, responses, and errors for the supported web service protocols.
- Documents the Amazon Fraud Detector commands that are available in the AWS CLI.
- Documents the reference information for all Amazon Fraud Detector resources and property types that AWS CloudFormation supports.
Other resources
- Use the Amazon Fraud Detector product page to gain an overall understanding of the product and if it meets your business use case.
- Find answers to frequently asked questions
- Use the blogs to create fraud detection solutions for your business use case.
- Ask questions or browse through the questions and answers to find solutions for issues with your specific use case.
PrivacySite termsCookie preferences
© 2025, Amazon Web Services, Inc. or its affiliates. All rights reserved.