Sampling from a distribution the 10% of values around the peak value in R

Question

I have a vector of 100 values.

bingo<-sample(0:5000, 100)

plot(density(bingo))

which.max(density(bingo)$y) # [1] 194

density(bingo)$x[194] # [1] 1507.085

I wanto to sample the 10% of the values around 1507.085, the peak of my distribution. Ho can I achieve it?

Many thanks for any advice!

Answer 1

It depends on what you mean by "around". Here are a couple of options:

set.seed(94)
bingo <- sample(0:5000, 100)
bingo <- sort(bingo)
bingo_mode <- density(bingo)$x[which.max(density(bingo)$y)]
idxMode <- match(TRUE, bingo > bingo_mode)

# to sample from 5 before and 5 after the peak (or the top/bottom decile if the
# peak is in the 95th/5th percentile)
if (idxMode < 5) {
  idxFrom <- 1
  idxTo <- 10
} else if (idxMode + 4 > 100) {
  idxFrom <- 91
  idxTo <- 100
} else {
  idxFrom <- idxMode - 5
  idxTo <- idxMode + 4
}

sample(bingo[idxFrom:idxTo], 1)
#> [1] 2557

# to sample from the 10 nearest the peak
if (idxMode < 10) {
  idxFrom <- 1
  idxTo <- 20
} else if (idxMode + 9 > 100) {
  idxFrom <- 81
  idxTo <- 100
} else {
  idxFrom <- idxMode - 10
  idxTo <- idxMode + 9
}

sample(bingo[idxFrom:idxTo][order(abs(bingo_mode - bingo[idxFrom:idxTo]))[1:10]], 1)
#> [1] 2678