Distribution-free inference for regression: discrete, continuous, and in between

Yonghoon Lee, Rina Foygel Barber

In data analysis problems where we are not able to rely on distributional assumptions, what types of inference guarantees can still be obtained? Many popular methods, such as holdo…