A team builds a model to predict whether a support ticket will escalate at creation time. The dataset includes the final resolution code, and duplicate copies of some tickets appear in both training and test data. Which changes improve the validity of the evaluation? Select 2 answers.

1. Keep the final resolution code because any field that improves test accuracy is suitable for deployment.
2. Move every difficult test example into the training set until the test score reaches the target.
3. Exclude features that become available only after the prediction time, such as the final resolution code.
4. Separate duplicate or closely related ticket records so the held-out set does not contain copies of training examples.
5. Train and test on the complete same dataset to eliminate the effects of sampling.