tidyst_ksupp: Tidy and geospatial kernel support estimate

In eks: Tidy and Geospatial Kernel Smoothing

Tidy and geospatial kernel support estimate

Description

Tidy and geospatial versions of a kernel support estimate for 2-dimensional data.

Usage

tidy_ksupp(data, cont=95, convex_hull=TRUE, ...)
st_ksupp(x, cont=95, convex_hull=TRUE, ...)

Arguments

data	tidy kernel density estimate (output from tidy_kde)
x	spatial kernel density estimate (output from st_kde)
cont	scalar contour level. Default is 95.
convex_hull	flag to compute convex hull of contour region. Default is TRUE.
...	other parameters in ks::ksupp function

Details

The kernel support estimate is considered to be the cont% probability contour of the kernel density estimate, with an additional convex hull calculation if convex_hull=TRUE. For details of the computation of the kernel support estimate, see ?ks::ksupp.

Value

The output from *_ksupp have the same structure as the kernel density estimate from *_kde, except that x,y indicate the boundary of the density support estimate (if convex.hull=TRUE) or the grid points inside the density support (if convex.hull=FALSE), rather than the complete grid points themselves.

For st_kdr, the density support estimate is stored as a (multi)polygon sf object.

Examples

## tidy density support estimate
library(ggplot2)
data(crabs, package="MASS")
crabs2 <- dplyr::select(crabs, FL, CW)
t1 <- tidy_kde(crabs2)
t2 <- tidy_ksupp(t1)
ggplot(t1, aes(x=FL, y=CW)) + 
    geom_contour_filled_ks(cont=c(25,50,75,95), colour="grey50") +
    geom_polygon(data=t2, aes(linetype=label), fill=NA, colour=1) +
    colorspace::scale_fill_discrete_sequential() +
    scale_linetype_manual(values="dashed", name="Support\nconvex hull")

ggplot(t2, aes(x=FL, y=CW)) + 
    geom_polygon(data=t2, aes(colour=label), fill=NA, linetype="dashed") +
    geom_point_ks(data=t1, colour=3) + scale_colour_manual(values=1)

## geospatial density support estimate
data(wa)
data(grevilleasf)
hakeoides <- dplyr::filter(grevilleasf, species=="hakeoides")
s1 <- st_kde(hakeoides)
s2 <- st_ksupp(s1)
s3 <- st_ksupp(s1, cont=97.5)

## base R plot
xlim <- c(1.2e5, 1.1e6); ylim <- c(6.1e6, 7.2e6)
plot(wa, xlim=xlim, ylim=ylim)
plot(s1, cont=c(25,50,75,95,97.5), add=TRUE, border="grey50")
plot(sf::st_geometry(hakeoides), add=TRUE, pch=16, col=1)
plot(s2, add=TRUE, lty=2, lwd=2)
plot(s3, add=TRUE, lty=3, lwd=2)

## geom_sf plot
gs1 <- ggplot(s1) + geom_sf(data=wa, fill=NA) + ggthemes::theme_map()
gs1 + geom_sf(data=st_get_contour(s1, cont=c(25,50,75,95,97.5)),
    aes(fill=label_percent(contlabel)), col="grey50") +
    geom_sf(data=st_get_contour(s2), 
    aes(linetype=contlabel), fill=NA, colour=1) + 
    geom_sf(data=st_get_contour(s3), 
    aes(linetype=contlabel), fill=NA, colour=1) + 
    geom_sf(data=hakeoides, aes(colour=species)) +
    colorspace::scale_fill_discrete_sequential(palette="Heat2") +
    scale_colour_manual(values=c(1,1)) + 
    guides(colour=guide_legend(title="Locations")) +
    scale_linetype_manual(values=c("dashed","dotted"), 
    name="Support\nconvex hull") +
    coord_sf(xlim=xlim, ylim=ylim)

eks documentation built on May 29, 2024, 1:17 a.m.