For CRM provider IGT Mariposa, building predictive analytics models for gaming clients was a time-consuming, complex process that could only produce a few models a year.

That changed last year, however, when IGT started working with Chicago-headquartered Apollo Data Technologies. IGT Mariposa is now using the Apollo Predictive Modeling Platform to automate many of the time-intensive tasks associated with delivering complex predictive models, including the data preparation and evaluation of hundreds or thousands of model configurations.

The modeling platform gives IGT Mariposa the ability to quickly evaluate hundreds of models to identify the best-performing predictive model to provide customers, said Javier Saenz, vice president of strategy, IGT.

 “For our clients it means highly actionable customer intelligence; for example more relevant and focused targeted communications and promotions, based on extracting customer behavior from the predictive models that previously were difficult to automate.”

Saenz noted in an interview that in the first six months IGT Mariposa worked with Apollo, some 700 models were created, whereas, without the modeling platform, IGT Mariposa could produce only a handful in two years.

“It’s just a giant leap forward. This just puts us in a whole different category,” he said.

Where this is going to have dramatic impact is in the server-based gaming environment, said Jeff Kaplan, managing director of Apollo Data Technologies, which also has offices in Seattle.

 “What’s really exciting about this technology is it has made tremendous powerful impact on other industries with increased revenues, increased customer satisfaction,” Kaplan said. “We see a huge opportunity in the gaming space to take this technology and pull it into server-based gaming. It’s really taking predictive analytics and putting on steroids.”

So when a player sits down at a server-based gaming machine and sticks in a player loyalty card, the predictive analytics will enable game recommendations and casino offers in much the same manner as Amazon.com recommends similar books or CDs to a customer based on previous purchases. That means, he said, the casino can “recommend the right games and that right time, the right offers at the right time, the right denomination at the right time to increase customer satisfaction, which increases loyalty, which is going to increase wallet share.”

For gaming operators, having such information will be key in the server-based world, Saenz said.

“In a world where you can change your floor on the fly, and dynamically redesign the gaming experience for your customers, we are building algorithms that are going to guide operators to the decisions they need to make,” he said.

That’s important, Saenz said, because “without the right information, they’re not going to have the confidence those changes to optimize the floor the way they could and the way they should.”

The modeling platform is metadata-driven, significantly reducing common rework in the areas of data preparation, algorithm selection, and model evaluation that is manually required by most data mining tools, according to Apollo Data Technologies.

It can run models and score records 24/7; will automatically identify the “optimal” model(s) to put into production or use for scoring; and can model tens of thousands of attributes and millions of records, the company said.

The platform is based on proprietary processes and algorithms used in a range of other industries, including retail and consumer goods, automotive, media and entertainment, telecommunications and healthcare. It is Windows-based and leverages the Microsoft SQL Server 2005 suite.