Changing assignment weights with time-based confounders
The Unofficial Google Data Science Blog
JULY 22, 2020
A/B testing isn’t simple just because data is big — the law of large numbers doesn’t take care of everything! Even with big data, A/B tests require thinking deeply and critically about whether or not the assumptions made match the data. An efficient bandit algorithm for realtime multivariate optimization."
95 
Let's personalize your content