試験

問題 9 / 65

For a new recommendation algorithm whose offline evaluation metrics have improved, which method is most appropriate for deciding whether to deploy it to production?

①Decide to deploy based solely on offline loss and accuracy.
②Decide based only on developers' intuition and reviews.
③Apply the results of academic papers directly to production.
④Decide whether to implement it based solely on local simulations.
⑤Conduct online A/B tests on real users and evaluate them using business metrics.

当サイトでは、ユーザー体験の向上を目的としてCookieを使用しています。サイトの利用を継続することで、Cookieの使用に同意したものとみなされます。