Turn per-block estimates into leave-one-out estimates

Source: R/resampling.R

This works for any statistics which, when computed across N blocks, are equal to the weighted mean of the statistics across the N blocks.

est_to_loo(arr, block_lengths = NULL)

Arguments

arr

3d array with blocked estimates, with blocks in the 3rd dimension

block_lengths

Optional block lengths. If NULL, will be parsed from 3rd dimnames in blocks

Value

A 3d array with leave-one-out estimates for jackknife. Dimensions are equal to those of arr.