Answer the question
In order to leave comments, you need to log in
How are these statistics problems solved?
You came up with a wizard that improves Yandex.Search and you want to test your idea in an A/B experiment. In just a day, users make 400 million requests, and your sorcerer is shown in 50% of requests. You have two selections: control and experimental (your sorcerer is shown only in the experimental). For the experiment, you have access to 100% of the request flow (50% for each sample). Determine how long the experiment should be in order for a 1% change in the metric "Share of clicked search results" with p-value = 0.05 to become statistically significant. Consider that before the experiment, the share of clicked search results was 10%. Describe your solution.
Answer the question
In order to leave comments, you need to log in
Didn't find what you were looking for?
Ask your questionAsk a Question
731 491 924 answers to any question