Innovatively Apply TreeNet to Predictive Analytics Abstract Visa Decision Sciences has utilized TreeNet®, a Salford-Systems product, during the last decade to meet an increasing demand for in-flight scoring and more accurate fraud detection. In this presentation, I will discuss how Visa Decision Sciences has creatively applied TreeNet to deliver a complicated suite of models with limited resources. The presentation will cover how we use TreeNet to: meet the real-time scoring; cut predictive model development cycle; eliminate hundreds of unimportant predictive variables; streamline development and production into one process with TreeNet PMML scoring capability; improve final model performance by segmenting data and assembling models; detect data issues early with TreeNet's feature ranking, and maintain stable score to fraud ratio relationship with uncovered high correlation between log(odds) and TreeNet raw score distribution.