New Machine Learning Model

By Kristin Hoyne Gomes, Director Decision Sciences May 22, 2015


Want to boost your decisioning effectiveness? Accertify is pleased to announce that we can now offer clients the ability to implement a new statistical model type built with machine learning to help reduce your review rate and improve your fraud capture rate.

This new machine learning model type leverages a proprietary algorithm from the Gradient Boosting (GBM) technique. Gradient boosting is a machine learning technique for regression problems, which produces a prediction model in the form of an ensemble of models. In the Accertify algorithm, the model ensemble is made of a series of decision trees. Each subsequent decision tree helps to optimize the decision of the prior tree, with the final result having a review by all trees.

In a recent client test run, GBM models have enabled improved decisioning over and above that of logistic regression models.

In addition, Accertify has implemented a new model reason capability that gives client fraud analysts clarity into the probability calculation from the model for each transaction.

This service is available through a consulting engagement with Accertify's Decision Science team.

Want to learn more, please contact Kristin Hoyne Gomes at or your Accertify Account Manager.

Posted in

About the Author

Kristin has over 17 years experience working in risk management and payments. At Accertify she leads a team responsible for improving risk decisioning.