Analyse the initial (first day's) results from the first A/B test to make sure the results don't have substantial biases.
Deadline for this is 1pm (Pacific) Wednesday 12th August 2015.
Analyse the initial (first day's) results from the first A/B test to make sure the results don't have substantial biases.
Deadline for this is 1pm (Pacific) Wednesday 12th August 2015.
Working on this too, btw. (There doesn't seem to be a way to assign a task to multiple users.)
To-do: exclude search sessions with prefix-search-only results and measure the effect size again.
@Deskana Sure! This is the presentation that Oliver and I did: https://github.com/wikimedia-research/SuggestItOff/blob/master/initial_analysis/presentation.Rmd