eks (version 1.1.1)

tidyst_kfs: Tidy and geospatial kernel feature significance

Description

Tidy and geospatial versions of kernel feature significance for 1- and 2-dimensional data.

Usage

tidy_kfs(data, ...)
st_kfs(x, ...)

Value

The output from tidy_kfs has the same structure as the kernel density estimate from tidy_kde, except that all values of estimate outside of the significant curvature regions are set to zero, and the label indicates whether the corresponding x,y point is inside a significant curvature region.

The output from st_kfs has a single contour, with contlabel=50, as a multipolygon which delimits significant curvature regions.

Arguments

data

data frame/tibble of data values

x

sf object with point geometry

...

other parameters in ks::kfs function

Details

A significant kernel curvature region consist of all points whose density curvature value is significantly different less than zero (i.e. forms a bump surrounding a local maximum). A Hochberg procedure is employed to control the significance level for multiple significance tests.

For details of the computation of the significant kernel curvature regions, see ?ks::kfs. The bandwidth matrix of smoothing parameters is computed as in ks::kdde(deriv_order=2).

Examples

## tidy significant curvature regions
library(ggplot2)
data(hsct, package="ks")
hsct <- dplyr::as_tibble(hsct)
hsct <- dplyr::filter(hsct, PE.Ly65Mac1>0 & APC.CD45.2>0)
hsct12 <- dplyr::filter(hsct, subject==12)   
hsct12 <- dplyr::select(hsct12, PE.Ly65Mac1, APC.CD45.2)
t1 <- tidy_kde(hsct12)
t2 <- tidy_kfs(hsct12)
gt <- ggplot(t2, aes(x=PE.Ly65Mac1, y=APC.CD45.2)) + 
    geom_contour_ks(data=t1, colour="grey50", cont=seq(10,90,by=10)) 
gt + geom_contour_filled_ks(aes(fill=after_stat(contregion)), colour=1) 
gt + geom_contour_ks(aes(colour=after_stat(contregion))) 

## geospatial significant curvature regions
data(wa)
data(grevilleasf)
hakeoides <- dplyr::filter(grevilleasf, species=="hakeoides")
s1 <- st_kfs(hakeoides)

## base R plot
xlim <- c(1.2e5, 1.1e6); ylim <- c(6.1e6, 7.2e6)
plot(wa, xlim=xlim, ylim=ylim)
plot(s1, add=TRUE)

## geom_sf plot
gs <- ggplot(s1) + geom_sf(data=wa, fill=NA) + ggthemes::theme_map() 
gs + geom_sf(data=st_get_contour(s1), aes(fill=contregion)) +
    coord_sf(xlim=xlim, ylim=ylim)