When presented with a sequence of observable variables {(x1,y1),…,(xn,yn)}, the conformal prediction method calculates a confidence set for yn+1 given xn+1 that remains valid for any finite sample size. This is achieved by assuming that the joint distribution of the data is permutation invariant. However, the computation of this set is often too complex for most regression problems…
For more information, refer to the paper details.
