Draws k seed points uniformly at random inside boundary, ignoring where
the observations are. Reach for this when the cells should cover the study
area evenly (so that sparsely sampled ground still gets its own cells and
is visibly under-sampled in the results) instead of concentrating
resolution where the data already are, which is what
voronoi_seeds_kmeans() does. It also suits a null or sensitivity
comparison: rerunning an analysis over several random seedings shows how
much of a result depends on one tessellation.
Arguments
- boundary
An sf or sfc polygonal object, the area the seeds are drawn in. Rows without a location, rows the repair leaves without a polygon, vertices at a latitude beyond 90 and mixed dimensions are handled as "Boundaries" in spatialkit-geometry says. The seeds are drawn as "Sampling inside the boundary" in
get_voronoi_seeds()describes, withkas the number of points asked for; the error for a boundary that would take too many draws says to ask for fewer seeds.- k
Integer; number of random seeds.
- set_seed
Optional integer RNG seed. Default
NULL: the seeds are drawn from the session's random-number stream, so consecutive calls give different seedings andset.seed()before a call makes it reproducible. Pass a number to get the same seeds whatever that stream holds; the caller's stream is then left as it was.
Value
An sf object of k random POINTs (fewer only in the degenerate
case above), with seed_id and method = "random" columns matching
get_voronoi_seeds(), and attr(, "cover"), the polygons of boundary:
build_tessellation() clips the cells to it (grown by 2 percent of its
diagonal) when no boundary is passed there, so the cells fill the area
the seeds were drawn in and not only the seeds' own hull.
Details
Seeds are drawn uniformly inside the polygon. When a draw falls short of k
it is topped up with further uniform draws from the same polygon, so the
result has exactly k seeds; only a geometry that still yields too few
after ten top-ups returns fewer, and that shortfall is logged.
Examples
library(sf)
bnd <- st_sf(geometry = st_sfc(st_polygon(list(rbind(
c(0, 0), c(100, 0), c(100, 100), c(0, 100), c(0, 0)
))), crs = 32632))
set.seed(1)
seeds <- voronoi_seeds_random(bnd, k = 10)
nrow(seeds) # 10
#> [1] 10
seeds
#> Simple feature collection with 10 features and 2 fields
#> Geometry type: POINT
#> Dimension: XY
#> Bounding box: xmin: 6.178627 ymin: 17.65568 xmax: 94.46753 ymax: 99.19061
#> Projected CRS: WGS 84 / UTM zone 32N
#> geometry seed_id method
#> 1 POINT (26.55087 20.59746) 1 random
#> 2 POINT (37.21239 17.65568) 2 random
#> 3 POINT (57.28534 68.70228) 3 random
#> 4 POINT (90.82078 38.41037) 4 random
#> 5 POINT (20.16819 76.98414) 5 random
#> 6 POINT (89.83897 49.76992) 6 random
#> 7 POINT (94.46753 71.76185) 7 random
#> 8 POINT (66.07978 99.19061) 8 random
#> 9 POINT (62.9114 38.00352) 9 random
#> 10 POINT (6.178627 77.74452) 10 random
# Another call is another seeding; set_seed pins one.
identical(st_coordinates(voronoi_seeds_random(bnd, k = 10, set_seed = 7)),
st_coordinates(voronoi_seeds_random(bnd, k = 10, set_seed = 7)))
#> [1] TRUE