Lisa Maag

Finding the Needle in the Haystack: GfK’s Journey from Manual Checks to Automated Quality Tracking


November 16, 2020


4:55 pm


In order to derive important business value for our clients we need high quality data and good performing models. To ensure those standards in production a close monitoring is essential. However, a manual inspection of the data and the related models is due to their complexity and limited capacity not always possible. GfK’s Auto-matching initiative supervises a data science solution to automatically integrate incoming retailer data in our Point of Sales panel. With about 500.000 stores, 20 million products and 50 different Auto-matching models neither the data nor the models can be monitored manually anymore. This talk is about the refinement of our setup to an automated monitoring system which supports us in tracking the quality of our outcome. Depending on the impact of model failure, this system ranges from long-term monitoring, over automated alters and notifications to a safety-net preventing the worst case without manual interference.

Ready to attend?

Register now! Join your peers.

Register nowView agenda

Knowledge is everything!
Sign up for our newsletter to receive:

  • an extra 10% off your ticket!
  • insights, interviews, tips, news, and much more about Predictive Analytics World
  • price break reminders