represampling_factor_bootstrap
resamples partitions defined by a
factor variable. This can be used for non-overlapping block bootstraps and
similar.
represampling_factor_bootstrap(data, fac, repetition = 1, nboot = -1, seed1 = NULL, oob = FALSE)
data |
|
---|---|
fac | defines a grouping or partitioning of the samples in |
repetition | numeric vector: cross-validation repetitions
to be generated. Note that this is not the number of repetitions,
but the indices of these repetitions. E.g., use |
nboot | number of bootstrap replications used for generating the
bootstrap training sample ( |
seed1 |
|
oob | if |
nboot
refers to the number of groups
(as defined by the factors) to be drawn with replacement from the set of
groups. I.e., if fac
is a factor variable, nboot
would normally
not be greater than nlevels(fac)
, nlevels(fac)
being the
default as per nboot = -1
.
represampling_disc_bootstrap, represampling_tile_bootstrap
data(ecuador) # a dummy example for demonstration, performing bootstrap # at the level of an arbitrary factor variable: parti <- represampling_factor_bootstrap(ecuador, factor(floor(ecuador$dem / 100)), oob = TRUE) # plot(parti,ecuador) # using the factor bootstrap for a non-overlapping block bootstrap # (see also represampling_tile_bootstrap): fac <- partition_tiles(ecuador, return_factor = TRUE, repetition = c(1:3), dsplit = 500, min_n = 200, rotation = 'random', offset = 'random') parti <- represampling_factor_bootstrap(ecuador, fac, oob = TRUE, repetition = c(1:3)) # plot(parti, ecuador)