An interviewer may ask you to explain a modeling choice, then ask you to show that your code follows it. These original tasks combine those two demands. You will identify the output type, build aligned feature and target objects, split observations, fit a baseline, and report the requested metric. The fictional scenarios are teaching exercises, not claims about questions used by named employers.
Keep the answer focused on what the supplied evidence can support. An impressive training score does not establish generalization, and an accuracy result does not automatically describe positive-class coverage. Some questions specify an exact split or estimator so another person can reproduce the answer. There are no new techniques in this set. Use the smallest complete calculation that answers the question and leaves the test partition intact.