Applies [summarize_by()] to each block. Explicit entity correspondence is preserved only when the entity key remains among the grouping variables. Otherwise the result is intentionally uncontracted with respect to rows. Masked blocks require an explicit coordinate-wise `aggregate` rule.
Usage
# S3 method for class 'hyperdesign'
summarize_by(x, ..., sfun = colMeans, extract_data = FALSE, aggregate = NULL)Examples
md <- multidesign(
matrix(1:8, nrow = 4),
data.frame(entity = c("A", "A", "B", "B")),
cells = matrix(c(TRUE, TRUE, TRUE, FALSE, TRUE, TRUE, FALSE, FALSE),
nrow = 4)
)
hd <- hyperdesign(list(one = md))
out <- summarize_by(hd, entity, aggregate = "mean")
cell_mask(out[[1]])
#> [,1] [,2]
#> group 1 TRUE TRUE
#> group 2 TRUE FALSE