sim_calculate_ij calculates similarities given pairs of rows.
sim_calculate_ij(
population,
index,
method = NULL,
annotation_prefix = "Metadata_",
...
)data.frame with annotations (a.k.a. metadata) and observation variables.
data.frame with at least two columns id1 and
id2 specifying rows of population, and an optional attribute
metric_metadata$method, which is a character string specifying
implemented. Preserve the diagonal entries when constructing index
optional character string specifying method for
to calculate similarity. method, if specified, overrides
attr(index, "metric_metadata")$method.
optional character string specifying prefix for annotation columns.
arguments passed downstream for parallel processing.
data.frame which is the same as index, but with a new
column sim containing similarities, and with the diagonals filtered out.
suppressMessages(suppressWarnings(library(magrittr)))
population <- tibble::tribble(
~Metadata_group, ~x, ~y, ~z,
1, -1, 5, -5,
2, 0, 6, -4,
3, 7, -4, 3,
4, 14, -8, 6
)
n <- nrow(population)
index <-
expand.grid(id1 = seq(n), id2 = seq(n), KEEP.OUT.ATTRS = FALSE)
matric::sim_calculate_ij(population, index, method = "cosine")
#> id1 id2 sim
#> 1 2 1 0.9709195
#> 2 3 1 -0.6836729
#> 3 4 1 -0.6836729
#> 4 1 2 0.9709195
#> 5 3 2 -0.5803433
#> 6 4 2 -0.5803433
#> 7 1 3 -0.6836729
#> 8 2 3 -0.5803433
#> 9 4 3 1.0000000
#> 10 1 4 -0.6836729
#> 11 2 4 -0.5803433
#> 12 3 4 1.0000000
attr(index, "metric_metadata") <- list(method = "cosine")
matric::sim_calculate_ij(population, index)
#> id1 id2 sim
#> 1 2 1 0.9709195
#> 2 3 1 -0.6836729
#> 3 4 1 -0.6836729
#> 4 1 2 0.9709195
#> 5 3 2 -0.5803433
#> 6 4 2 -0.5803433
#> 7 1 3 -0.6836729
#> 8 2 3 -0.5803433
#> 9 4 3 1.0000000
#> 10 1 4 -0.6836729
#> 11 2 4 -0.5803433
#> 12 3 4 1.0000000