Leave-K-Out Cross Validation and Bias Correction #21
Description
Currently, only leave-one-out CV is supported; leaving out more observations might be a good idea. If we do this, we should probably introduce bias corrections for cross-validation, since leave-k-out with k >> 1 can introduce large biases.