数据行元素的成对比较

Question

我想在数据帧中找到所有公共元素的数量。

name           members
x1            A,B,N,K,Y,G
x2            J,L,M,N,T
x3            G,H,S,J,D,F
x4            J,K,H,F,H,D,L

name         common       name
x1                   6               x1
x1                   2               x2
x1                     -             x3
x1                      -            x4
x2                       -           x1
x2                5        -          x2
x2                         -         x3
x2                          -        x4
x3                           -       x1
x3                            -      x2
x3                   6          -     x3
x3                              -    x4
x4                               -   x1
x4                                -  x2
x4                                 - x3
x4                  7                -x4

Answer 1

我相信下面的代码可以完成问题所要求的内容。但是请注意，我发现它很复杂，有两个merge指令，也许其他人会找到一个更简单的解决方案。

fun <- function(DF){
  ex <- expand.grid(Var2 = DF[['name']], name = DF[['name']])[2:1]
  members <- as.character(DF[['members']])
  merge(DF, ex)
}

tmp <- merge(df1, fun(df1))
o <- order(tmp[[3]])
tmp$members2 <- tmp$members[o]

tmp$common <- apply(tmp[c(2, 4)], 1, function(x){
  y1 <- unlist(strsplit(as.character(x[1]), ","))
  y2 <- unlist(strsplit(as.character(x[2]), ","))
  length(intersect(y1, y2))
})

res <- tmp[c(1, 5, 3)]
names(res)[3] <- "name2"

head(res)
#  name common name2
#1   x1      6    x1
#2   x1      1    x2
#3   x1      1    x3
#4   x1      1    x4
#5   x2      1    x1
#6   x2      5    x2

最后清理一下。

rm(tmp)

数据.

df1 <- read.table(text = "
name           members
x1            A,B,N,K,Y,G
x2            J,L,M,N,T
x3            G,H,S,J,D,F
x4            J,K,H,F,H,D,L
", header = TRUE)

Answer 2

1) dplyr/tidyr对于每一行，使用separate_rows为每个成员创建一个单独的行，并通过members将其连接到自己。然后计算计数并完成它。

library(dplyr)
library(tidyr)

DF %>%
  separate_rows(members) %>%
  distinct %>%
  inner_join(., ., by = "members") %>%
  count(name.x, name.y) %>%
  complete(name.x, name.y)

给予：

# A tibble: 16 x 3
   name.x name.y     n
   <chr>  <chr>  <int>
 1 x1     x1         6
 2 x1     x2         1
 3 x1     x3         1
 4 x1     x4         1
 5 x2     x1         1
 6 x2     x2         5
 7 x2     x3         1
 8 x2     x4         2
 9 x3     x1         1
10 x3     x2         1
11 x3     x3         6
12 x3     x4         4
13 x4     x1         1
14 x4     x2         2
15 x4     x3         4
16 x4     x4         6

2) Base 创建一个函数，用于计算两个成员组件之间的交叉点数。然后使用outer将其应用于每一对，并转换为data.frame。

Scan <- function(x) scan(text = x, what = "", sep = ",", quiet = TRUE)
countSame <- function(x, y) length(intersect(Scan(x), Scan(y)))
x <- setNames(DF$members, DF$name)
as.data.frame.table(outer(x, x, Vectorize(countSame)))

给予：

   Var1 Var2 Freq
1    x1   x1    6
2    x2   x1    1
3    x3   x1    1
4    x4   x1    1
5    x1   x2    1
6    x2   x2    5
7    x3   x2    1
8    x4   x2    2
9    x1   x3    1
10   x2   x3    1
11   x3   x3    6
12   x4   x3    4
13   x1   x4    1
14   x2   x4    2
15   x3   x4    4
16   x4   x4    6

尽管上面询问的是data.frame表单，但您可能更喜欢2d表，它可以通过从最后一行代码中省略as.data.frame.table来生成。

   x1 x2 x3 x4
x1  6  1  1  1
x2  1  5  1  2
x3  1  1  6  4
x4  1  2  4  6

(2)只有两条线长的变异，可以通过对成员进行串分裂，然后利用外部计算成对相交的长度来形成变化。最后，我们将转换成一个数据帧。可以通过省略as.data.frame.table再次形成2d表。)

x <- with(DF, setNames(strsplit(members, ","), name))
as.data.frame.table(outer(x, x, Vectorize(function(x, y) length(intersect(x, y)))))

给予：

   Var1 Var2 Freq
1    x1   x1    6
2    x2   x1    1
3    x3   x1    1
4    x4   x1    1
5    x1   x2    1
6    x2   x2    5
7    x3   x2    1
8    x4   x2    2
9    x1   x3    1
10   x2   x3    1
11   x3   x3    6
12   x4   x3    4
13   x1   x4    1
14   x2   x4    2
15   x3   x4    4
16   x4   x4    6

备注

Lines <- "name           members
x1            A,B,N,K,Y,G
x2            J,L,M,N,T
x3            G,H,S,J,D,F
x4            J,K,H,F,H,D,L"
DF <- read.table(text = Lines, header = TRUE, as.is = TRUE)