Re:The multiple comparisons problem
by rfergie on Wednesday January 07, 2015 @11:36AM
(#8851463)
You should look at multi-armed bandits - this approach is pretty much what you have reached in your conclusion and a lot of work has been done on them
[ Parent
| on HN ]
Funnily enough, my starting point to get into A/B testing was research on multiarmed bandits.
Update: I should add I think the issues are more complex than simply using, say, Thompson sampling.