AWS Certification Exam AI Practitioner (Practice Questions) set3

問題 64 / 65

When you suspect data leakage because only some departments show large performance differences between training and evaluation data, which validation design should you check first?

Create an ensemble of models to improve stability.
Deploy the model to production immediately in accordance with the policy.
Modify the evaluation metric and re-evaluate.
Delete inference logs to clean the dataset.
Validate using cross-validation and external holdouts that take department and time into account.

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。