Advanced analytics, machine learning and logistical regression – traditional methods used to calculate credit scores today. Enter deep neural networks, the mysterious black box that was once thought uninterpretable. While modern techniques such as this promise more accurate results, the challenge was in interpreting the results in order to maintain regulatory compliance.

In a recent interview with Forbes, Peter Maynard, Senior Vice President of Global Analytics at Equifax discusses the advances Equifax has made in decoding neural networks. “We developed a mathematical proof that shows that we could generate a neural net solution that can be completely interpretable for regulatory purposes,” explains Maynard.

Not only can incorporating neural networks improve automation and speed model development time, it also enables the use of  larger data sets, including trended data which enables identification of patterns in the data set over a period of time, up to 72 months of data for Equifax.

Perhaps most importantly, Maynard notes “the neural net has improved the predictive ability of the model by up to 15%.” And this translates to greater access to credit. In fact, not only does Equifax have the ability to use up to 72 months of trended data in its calculations, it can also tap into alternative sources of data to further expand existing data sets being analyzed today. And Equifax can open that data store up to its customers via Equifax Ignite Direct. Customers can have direct access to data and tools within a single, secure environment for definition, development, validation and operational deployment of models and scores.

To read the full article: Equifax And SAS Leverage AI And Deep Learning To Improve Consumer Access To Credit Learn more about Equifax Ignite Direct Learn more about Trended Data solutions at Equifax Learn more about Alternative Data solutions at Equifax