Using MOE, the Metric Optimization Engine, to Optimize an A/B Testing Experiment Framework
-
Dr. Scott Clark, Ad Targeting Engineer
- Oct 9, 2014
A/B Testing Experiment Frameworks and MOE We recently open sourced MOE, the Metric Optimization Engine, a machine learning tool for solving global, black box optimization problems. An example application for such a system is optimally running online A/B experiments. A/B testing segments the users that come to a site into buckets, or cohorts, and show different versions of the site to different cohorts of users. One can show 50% of users one version of a site (version A) and 50% of users another version of a site (version B). After some amount of time we can see which version of...