Tables of a grouped Markov order test

The per-group test tables stacked into one data.frame, with a group column naming the group each row came from. This is the supported way to read a net_markov_order_group; the print method shows each group in turn but a comparison across groups needs one table.

Usage

# S3 method for class 'net_markov_order_group'
hg_get(x, what = c("orders", "null"), ..., top = NULL)

Arguments

x

A net_markov_order_group, as returned by markov_order() on a netobject_group.

what

"orders" (default) for one row per group and order, or "null" for the permutation null draws, one row per group, order and replicate.

...

Unused.

top

Keep only the first top rows, applied last.

Value

A base data.frame. For what = "orders", one row per group and order, with group first and then the columns of hg_get.net_markov_order(). For what = "null", one row per group, order and replicate with columns group, order, replicate, g2. Returns a zero-row frame with those columns when the group is empty.

Examples

seqs <- list(c("a", "b", "a", "c"), c("b", "a", "c", "a"))
fit <- markov_order(seqs, max_order = 2L, n_perm = 20L, seed = 1L)
hg_get(fit)
#>   order log_likelihood      aic      bic df       g2   p_value p_asymptotic
#> 1     0      -8.317766 20.63553 20.79442 NA       NA        NA           NA
#> 2     1      -3.988984 13.97797 14.21629  4 8.317766 0.3333333   0.08060755
#> 3     2      -3.178054 12.35611 12.59443  0 0.000000        NA           NA
#>   significant
#> 1          NA
#> 2       FALSE
#> 3       FALSE