在数据框中按多个组标记唯一值

Question

我在 R 中有一个大型数据框，其中用户的任务是描述场景中的对象。每个场景我需要唯一的 3 个用户，但是有些场景被描述了 3 次以上。我试图保留前 3 个唯一用户并删除其余用户。

玩具数据（真实数据集有更多的行和列）

user <- c("A", "A", "A", "B", "B", "C", "C", "D", "E", "E", "F", "F", "F")
scene <- c("library", "library", "library", "park", "park", "library", "library", "park", "library", "library", "library", "library", "library")
object <- c("book", "book", "lamp", "dog", "cat", "book", "lamp", "dog", "desk", "desk", "book", "lamp", "lamp")
index <- c(1, 2, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 2)
dat <- data.frame(user, scene, object, index)

user     scene      object      index
A        library    book        1
A        library    book        2
A        library    lamp        1
B        park       dog         1
B        park       cat         1
C        library    book        1
C        library    lamp        1
D        park       dog         1
E        library    desk        1
E        library    desk        2
F        library    book        1
F        library    lamp        1
F        library    lamp        2
...      ...        ...         ...

例如，这里A、B、C是第一个描述场景library的用户。所以现在不需要F的描述了。我的主要问题是，虽然我可以获得唯一身份用户的总体数量，但我不知道如何将它们标记为 1、2、3 等以切断超过 3 的值。

期望的输出

user     scene      object      index   count
A        library    book        1       1
A        library    book        2       1
A        library    lamp        1       1
B        park       dog         1       1
B        park       cat         1       1
C        library    book        1       2
C        library    lamp        1       2
D        park       dog         1       2
E        library    desk        1       3
E        library    desk        2       3

这很有用，但只能按一列分组，所以我无法在此处应用它：

Answer 1

对于每个 user，您可以使用 match 创建一个 count 变量，然后 filter 输出值直到 count <= 3 :

library(dplyr)

dat %>%
  group_by(scene) %>%
  mutate(count = match(user, unique(user))) %>%
  filter(count <= 3)

#   user  scene   object index count
#   <chr> <chr>   <chr>  <dbl> <int>
# 1 A     library book       1     1
# 2 A     library book       2     1
# 3 A     library lamp       1     1
# 4 B     park    dog        1     1
# 5 B     park    cat        1     1
# 6 C     library book       1     2
# 7 C     library lamp       1     2
# 8 D     park    dog        1     2
# 9 E     library desk       1     3
#10 E     library desk       2     3

data.table 中的相同内容是：

library(data.table)
setDT(dat)[, count := match(user, unique(user)), scene]
dat[count <= 3]

和基础 R :

dat$count <- with(dat, ave(user, scene, FUN = function(x) match(x, unique(x))))
subset(dat, count <= 3)

在数据框中按多个组标记唯一值

Label unique values by multiple groups in dataframe

r

dataframe

data-wrangling