使用 dplyr 函数循环？

Question

我有一个大型数据集，想查看各种图书馆（所有者）和拥有的材料（对象）之间的关系。到目前为止，我一直在手动执行此操作，通过传播和收集来计算每个所有者之间的重叠。

有没有办法循环这种类型的函数，这样我就不必重复 n 次了？

这是一个数据样本（已经传播），工作确实将其收集回来。

library(dplyr)

object <- c(1:10)
A <- sample(0:1, 10, replace = TRUE)
B <- sample(0:1, 10, replace = TRUE)
C <- sample(0:1, 10, replace = TRUE)
D <- sample(0:1, 10, replace = TRUE)

df <- data.frame(object, A, B, C, D)

dfA <- df %>% filter(A == 1)
dfA$owner1 <- "A"
dfA <- dfA %>% gather(owner2, overlap, A:D, factor_key = TRUE)
dfA <- dfA %>% filter(overlap != 0)

dfB <- df %>% filter(B == 1)
dfB$owner1 <- "B"
dfB <- dfB %>% gather(owner2, overlap, A:D, factor_key = TRUE)
dfB <- dfB %>% filter(overlap != 0)

dfC <- df %>% filter(C == 1)
dfC$owner1 <- "C"
dfC <- dfC %>% gather(owner2, overlap, A:D, factor_key = TRUE)
dfC <- dfC %>% filter(overlap != 0)

dfD <- df %>% filter(D == 1)
dfD$owner1 <- "D"
dfD <- dfD %>% gather(owner2, overlap, A:D, factor_key = TRUE)
dfD <- dfD %>% filter(overlap != 0)

df2 <- rbind(dfA, dfB) %>% rbind(dfC) %>% rbind(dfD) 

结果允许您计算每个所有者拥有多少个对象（当它是 owner1 和 owner2 时），以及所有者之间的重叠。

谢谢

Answer 1

我使用 tidyverse 想出了以下内容。有点密集，所以加了cmets：

pmap(df, \(...) {                           # iterate through rows
  c(...)[-1] %>%                            # drop first column
    .[. == 1] %>%                           # columns with value 1
    names() %>%                             # owners (A, B, ..)
    (\(xs) expand.grid(xs, xs)) %>%         # generate pairs of owners
    mutate(object = c(...)[1], .before = 1) # add object column
}) %>%
  bind_rows()                               # combine

返回（缩写）：

  object Var1 Var2
1      1    A    A
2      1    C    A
3      1    A    C
4      1    C    C
5      2    A    A
6      2    B    A
...

对于 R

pmap(df, function(...) {
  c(...)[-1] %>%
    .[. == 1] %>%
    names() %>%
    (function(xs) expand.grid(xs, xs)) %>%
    mutate(object = c(...)[1], .before = 1)
}) %>%
  bind_rows()