An ML engineer peeks at the test data and observes that the predicted variable "y" is quadratic in terms of the observed variable "x". So, he chooses model y = ax2 + bx + c.Is he right in doing so?
(A) This is okay since we are not estimating the parameters a,b,c from the test data
(B) This is not okay since test data should not be used for inference.
(C) This is okay since test data should not be used for inference.
(C) This is not okay since we are not estimating the parameters a,b,c from the test data.
Comments
Post a Comment