Stratified random sample
Creates a stratified random sample of an sp class object
stratified.random(x, strata, n = 10, reps = 1, replace = TRUE)
x |
sp class SpatialDataFrame object (point, polygon, line, pixel) |
strata |
Column in @data slot with stratification factor |
n |
Number of random samples |
reps |
Number of replicates per strata |
replace |
Sampling with replacement (TRUE|FALSE) |
sp SpatialDataFrame object (same as input feature) containing random samples
If replace=FALSE features are removed from consideration in subsequent replicates. Conversely, if replace=TRUE, a feature can be selected multiple times across replicates. Not applicable if rep=1.
Depends: sp
Jeffrey S. Evans <jeffrey_evans@tnc.org>
Hudak, A.T., N.L. Crookston, J.S. Evans, M.J. Falkowski, A.M.S. Smith, P. Gessler and P. Morgan. (2006) Regression modelling and mapping of coniferous forest basal area and tree density from discrete-return lidar and multispectral satellite data. Canadian Journal of Remote Sensing 32: 126-138.
require(sp)
data(meuse)
coordinates(meuse) <- ~x+y
# Create stratified variable using quartile breaks
x1 <- cut(meuse@data[,'cadmium'], summary(meuse@data[,'cadmium'])[-4],
include.lowest=TRUE)
levels(x1) <- seq(1,nlevels(x1),1)
x2 <- cut(meuse@data[,'lead'], summary(meuse@data[,'lead'])[-4],
include.lowest=TRUE)
levels(x2) <- seq(1,nlevels(x2),1)
meuse@data <- cbind(meuse@data, STRAT=paste(x1, x2, sep='.') )
# 2 replicates and replacement
ssample <- stratified.random(meuse, strata='STRAT', n=2, reps=2)
# 2 replicates and no replacement
ssample.nr <- stratified.random(meuse, strata='STRAT', n=2, reps=2,
replace=FALSE)
# n=1 and reps=10 for sequential numbering of samples
ssample.ct <- stratified.random(meuse, strata='STRAT', n=1, reps=10,
replace=TRUE)
# Counts for each full strata (note; 2 strata have only 1 observation)
tapply(meuse@data$STRAT, meuse@data$STRAT, length)
# Counts for each sampled strata, with replacement
tapply(ssample@data$STRAT, ssample@data$STRAT, length)
# Counts for each sampled strata, without replacement
tapply(ssample.nr@data$STRAT, ssample.nr@data$STRAT, length)
# Counts for each sampled strata, without replacement
tapply(ssample.ct@data$STRAT, ssample.ct@data$STRAT, length)
# Plot random samples colored by replacement
ssample@data$REP <- factor(ssample@data$REP)
spplot(ssample, 'REP', col.regions=c('red','blue'))Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.