August 30, 2011

FICO announced the general availability of FICO Model Builder 7.2, which enables analytic developers to discover, design, and deploy predictive analytics for high-volume decisions.

It includes new capabilities to develop and deploy optimized suites of segmented ensemble models, which marry decision trees and scorecards to capitalize on the predictive interactions across customer sub segments. These advancements enable companies to improve the productivity of analytic staff, and make more profitable decisions based on more precise risk estimates.

Specifically, FICO Model Builder 7.2 combines advanced analytic science with business expertise—automating the discovery of optimal segmentation schemes and accelerating the refinement of subpopulation models to compare different ways of segmenting the population. It features a revamped segmentation discovery algorithm that improves runtime, scalability, and accuracy, while giving analysts control to manage and direct their search for the most powerful segmentation scheme.

In addition, the product automates deployment of the complete ensemble model, including variable generation logic, model selection, and reason codes assignment, for implementation within FICO applications or other systems. All of these features lead to more precise risk estimates, increased staff productivity, and improved organizational profitability.

“Financial institutions have long valued segmented ensembles for their ability to capture predictive analytics in a transparent, easily understood form,” said James Taylor, CEO of Decision Management Solutions, and a leading expert in the use of predictive analytics. “This structured combination of decision trees and scorecards…makes it easy to find the best segmentation and quickly build the most effective ensemble.”

“In the data mining community, we’re all excited about the advances from number-crunching ensemble modeling techniques,” said Andrew Flint, senior director of product management at FICO. “To take full advantage of ensembles for the closely managed decisions in lending, underwriting and pricing, creditors and insurers require great transparency and control over the ensemble modeling process. Our customers strongly value the deep user controls that Model Builder 7.2 brings to these modern analytic algorithms.”


http://www.fico.com/en/Company/News/Pages/08-18-2011.aspx