新しい (より簡単な) 解決策: Ari B. Friedman のパッケージのshift
関数を使用するTaRifx
tt <- sample(people)
lapply(seq_len(length(tt))-1, function(x) shift(tt, x)[1:3])
# if you don't want it to be ordered, just add a sample(.)
lapply(seq_len(length(tt))-1, function(x) sample(shift(tt, x)[1:3]))
# [[1]]
# [1] "Bob" "Frank" "Betty"
#
# [[2]]
# [1] "Frank" "Betty" "Joe"
#
# [[3]]
# [1] "Betty" "Joe" "Will"
#
# [[4]]
# [1] "Joe" "Will" "Bob"
#
# [[5]]
# [1] "Will" "Bob" "Frank"
古い解決策(アイデア):
私はこのように行きます。基本的にはsample
「人」になったら、1,2,3, 2,3,4, 3,4,5, 4,5,1 といつでも行けます。では、そうしましょう。つまり、これらのインデックスを生成してから、人々をサンプリングしてトリプレットを取得します。
# generate index
len <- length(people)
choose <- 3 # at a time
idx <- outer(seq(choose), seq(choose+2)-1, '+')
# [,1] [,2] [,3] [,4] [,5]
# [1,] 1 2 3 4 5
# [2,] 2 3 4 5 6
# [3,] 3 4 5 6 7
# sample people
tt <- sample(people)
# [1] "Joe" "Will" "Bob" "Frank" "Betty"
max.idx <- 2*choose + 1
tt[(len+1):max.idx] <- tt[seq(max.idx-len)]
# [1] "Joe" "Will" "Bob" "Frank" "Betty" "Joe" "Will"
tt[idx]
# [1] "Joe" "Will" "Bob" "Will" "Bob" "Frank" "Bob" "Frank" "Betty" "Frank"
# [15] "Betty" "Joe" "Betty" "Joe" "Will"
split(tt[idx], gl(ncol(idx), nrow(idx)))
# $`1`
# [1] "Joe" "Will" "Bob"
#
# $`2`
# [1] "Will" "Bob" "Frank"
#
# $`3`
# [1] "Bob" "Frank" "Betty"
#
# $`4`
# [1] "Frank" "Betty" "Joe"
#
# $`5`
# [1] "Betty" "Joe" "Will"
これで、これをすべて関数に入れることができます。
my_sampler <- function(x, choose) {
len <- length(x)
idx <- outer(seq(choose), seq(choose+2)-1, '+')
sx <- sample(x)
max.idx <- 2*choose + 1
sx[(len+1):max.idx] <- sx[seq(max.idx-len)]
split(sx[idx], gl(ncol(idx), nrow(idx)))
}
# try it out
my_sampler(people, 3)
my_sampler(people, 4) # 4 at a time
# if you want this and want a non-ordered solution, wrap this with `lapply` and `sample`
lapply(my_sampler(people, 3), sample)