Amazon Fraud Detector, a fully managed fraud prevention service offered natively within Amazon Web Services (AWS) now includes model variable importance values, the company announced in a press release.
Model variable importance values provides merchants with a ranked list of model inputs based on their relative importance in the machine learning (ML) models used by Amazon Fraud Detector. The information is intended to both provide merchants with more information about how the model works, and give them more insight into its performance.
“With model variable importance, Fraud Detector now provides a ranked list of model inputs based on their relative importance to the model’s performance. This information helps customers better understand their ML models and makes it easier to iteratively improve model performance,” the statement said.
Amazon Fraud Detector also gives merchants some customization options. For example, if a merchant notices that a certain IP address is the source of many fraudulent attempts, they will be able to add additional sub-attributes of IP addresses as factors in the algorithm.
Amazon also noted in its release that the feature is available wherever Amazon Fraud Detector is: US East (N. Virginia), US East (Ohio), US West (Oregon), Europe (Ireland), Asia Pacific (Singapore) and Asia Pacific (Sydney).