SAS: Breaking Down the Barriers Between Model Development and Deployment
Posted: 1 June 2017 | Source: SAS
Models are at the heart of all business decisions. As a result, analytics has become top of mind for the corporate C-suite. And increasingly, companies are looking to analytics as a key strategic differentiator to drive huge numbers of operational decisions every minute of every day.
But the reality is, due to processes that are often manual and ad hoc, it can take months to get a model implemented. Unfortunately, by the time models reach production, they may be out of sync with business needs. This can result in bad decisions; imagine a credit card company using a risk model that wasn’t up to date with the latest fraud schemes.
The main reason models take so long to implement is because companies produce and deploy them in two separate environments − business and IT. Analysts from the various business units build the models, and IT deploys them. Both groups rely on a different set of processes, programming codes and languages. The result is a lot of iteration and manual drudgery that makes it hard to react rapidly to changing business conditions.
If companies want to streamline model production, a better idea is to automate repeatable aspects of the job and migrate the models into a continuous decisionmaking process. The answer is an analytics factory.