A. Remove the outliers as suggested by Einstein and deploy the model.
B. Remove the outliers as suggested by Einstein, and verify using model metrics and story insights if the quality improved.
C. Discuss with the client if values below 0 and above 2,489 are so uncommon that they should perhaps be left out of the story.
D. Manually remove the sales below 0 (negative sales must be a data issue), but keep the large values (the more data, the better the model will be).

Suggest answer: BC

Reference:

No comments:

Please share to make the community better.