run_sampler_geo: Aggregated and overdispersed sampling for geographic...
In BrunoVilela/sampler: Aggregated and overdispersed sampling

Description Usage Arguments Details Value Author(s) See Also Examples

Generate aggregated or overdispersed sampling designs for any given coordinates. Results can be used to design experiments/samples, for resample proposes and data bias removal.

1 2	run_sampler_geo(x, n, alpha, dist.func = rdist.earth, n_start = 1, return_start = FALSE, starting = NULL)

`x`	`matrix` or `data.frame` indicating the coordinates (first column = longitude; second column = latitude). Row names should be given.
`n`	A positive integer number indicating the sample size.
`alpha`	Number indicating the strength of aggregation (if negative) or overdispersion (if positive). When alpha = 0 sample is random.
`dist.func`	A distance function to calculate coordinates distance. Default is `rdist.earth` from package `fields`.
`n_start`	Number of initial selected points. Default is one starting point.
`return_start`	if `TRUE` the starting point is returned.
`starting`	Character vector indicating the starting point (= to row names). If not provided, random starting value(s) is(are) selected.

The function uses the algorithm in run_sampler, but here it accepts a two column matrix of coordinates as input.

The function returns a vector indicating the selected rows. If return_start is TRUE, a list is returned with the first element being the Sampling_selection - selected sampling units - and Starting_points - selected starting point(s).

Bruno Vilela

run_sampler

require(sp)
require(maptools)
data(wrld_simpl)  # World map
Brazil <- wrld_simpl[wrld_simpl$NAME == "Brazil", ]  # Brazil (polygon)
coords <- slot(spsample(Brazil, 100, "regular"), "coords")
rownames(coords) <- paste0("t", 1:nrow(coords))
## Subsample 50%
### Overdispersed
selection.geo <- run_sampler_geo(x = coords, n = 10, alpha = 100, starting = "t10")
### Aggregated
selection.geo2 <- run_sampler_geo(x = coords, n = 10, alpha = -100, starting = "t10")
### Random
selection.geo3 <- run_sampler_geo(x = coords, n = 10, alpha = 0, starting = "t10")

## Plot
par(mfrow = c(1, 3), mar = c(1, 1, 15, 1))
plot(Brazil, main = "Overdispersed 50% sampling (red were selected)")
points(selection.geo, cex = 2, pch = 19, col = "red")
plot(Brazil, main = "Aggregated 50% sampling (blue were selected)")
points(selection.geo2, cex = 2, pch = 19, col = "blue")
plot(Brazil, main = "Random 50% sampling (green were selected)")
points(selection.geo3, cex = 2, pch = 19, col = "green")