mopa (version 1.0.1)

pseudoAbsences: Pseudo-absence data generation

Description

Pseudo-absence data generation at random or by k-means clustering inside a single background or a group of backgrounds (e.g. of different extent, backgroundRadius)

Usage

pseudoAbsences(xy, background, realizations = 1, exclusion.buffer = 0.0166,
  prevalence = 0.5, kmeans = FALSE, varstack = NULL)

Arguments

xy

Data frame or list of data frames with coordinates (each row is a point), this is, presence data

background

Matrix or list/s of matrixes with background coordinates in columns. Object derived from function backgroundGrid, OCSVMprofiling or backgroundRadius.

realizations

Integer. Number of realizations (default = 1).

exclusion.buffer

value of the minimum distance to be kept between presence data and pseudo-absence data. Default is 0.0166

prevalence

Proportion of presences against absences. Default is 0.5 (equal number of presences and absences)

kmeans

Logical. If FALSE (default) pseudo-absences are generated at random. If TRUE k-means clustering of the background is done and centroids are extracted as pseudo-absences.

varstack

RasterStack of variables for to compute the k-means clustering. Used if kmeans = TRUE.

Value

data frame or list/s of data frames

Details

Details. The application of this function could be preceded by the application of functions OCSVMprofiling and/or backgroundRadius in order to consider alternative methods for pseudo-absence data generation (see references).

References

Iturbide, M., Bedia, J., Herrera, S., del Hierro, O., Pinto, M., Gutierrez, J.M., 2015. A framework for species distribution modelling with improved pseudo-absence generation. Ecological Modelling. DOI:10.1016/j.ecolmodel.2015.05.018.

See Also

mopaTrain

Examples

# NOT RUN {
# SHORT EXAMPLE
## Load and prepare presence data
data(Q_pubescens)
presences <- Q_pubescens[sample(1:300, size = 100),]

## Define the spatial characteristics of the study area
r <- raster(nrows=50, ncols=50, xmn=-10, xmx=20, ymn=35, ymx=65, vals = rep(1, 50*50))

## Background of the whole study area
bg <- backgroundGrid(r)

## Generate pseudo-absences considering an unique background extent
RS_random <-pseudoAbsences(xy = presences, background = bg$xy, 
                           exclusion.buffer = 0.083*5, 
                           prevalence = -0.5, kmeans = FALSE)


# }
# NOT RUN {
# FULL WORKED EXAMPLE
## Load presence data
data(Oak_phylo2)

## Load climate data 
destfile <- tempfile()
data.url <- "https://raw.githubusercontent.com/SantanderMetGroup/mopa/master/data/biostack.rda"
download.file(data.url, destfile)
load(destfile, verbose = TRUE)

projection(biostack$baseline) <- CRS("+proj=longlat +init=epsg:4326")
r <- biostack$baseline[[1]]
## Background of the whole study area
bg <- backgroundGrid(r)

## Environmental profiling of the background
bg.profiled <- OCSVMprofiling(xy = Oak_phylo2, varstack = biostack$baseline, 
                              background = bg$xy)

## Generate pseudo-absences considering an unique background extent
RS_random <-pseudoAbsences(xy = Oak_phylo2, background = bg$xy, 
                           exclusion.buffer = 0.083*5, 
                           prevalence = -0.5, kmeans = FALSE)
RSEP_random <-pseudoAbsences(xy = Oak_phylo2, background = bg.profiled$absence, 
                             exclusion.buffer = 0.083*5, 
                             prevalence = -0.5, kmeans = FALSE)

## Background partition into different extents
bg.extents <- backgroundRadius(xy = Oak_phylo2, background = bg$xy, 
                               start = 0.166, by = 0.083*20, 
                               unit = "decimal degrees")

## Generate pseudo-absences considering different background extents
TS_random <-pseudoAbsences(xy = Oak_phylo2, background = bg.extents, 
                           exclusion.buffer = 0.083*10, 
                           prevalence = -0.5, kmeans = FALSE)


## with k-means clustering
TS_kmeans <-pseudoAbsences(xy = Oak_phylo2, background = bg.extents, 
                           exclusion.buffer = 0.083*5, 
                           prevalence = -0.5, kmeans = TRUE, 
                           varstack = biostack$baseline)
# }
# NOT RUN {
# }