r - 在 R 中 - 找到最小数量的单元格创建小于 n 的组

Question

我有一个包含多个分类列的数据框，具有不同数量的唯一条目。当我对所有列进行 group_by 和汇总时，存在小于 n 的组，其中 n 是例如2.例如:

> df
   A B  C
1  x z a1
2  x z a2
3  x z a1
4  x w a1
5  x w a2
6  y w a1
7  y u a2
8  y u a2
9  y u a1
10 y u a1

DF = df %>% group_by_at(c(1:3)) %>% count()

# A tibble: 7 x 4
# Groups:   A, B, C [7]
  A     B     C         n
  <chr> <chr> <chr> <int>
1 x     w     a1        1
2 x     w     a2        1
3 x     z     a1        2
4 x     z     a2        1
5 y     u     a1        2
6 y     u     a2        2
7 y     w     a1        1

查找哪些单元格创建了小于 n 的组并将其值替换为一个统一值(假设为“其他”)的最有效方法是什么，以便在此过程中创建的最小组的大小ñ？我需要对更大的数据集执行此过程。

Answer 1

有多种方法可以解决这个问题，例如，您只能将 B 中的所有 w 和 z 替换为其他。我能想到的最简单和最快的解决方案可能是使用 data.table，但这种方法是否有意义取决于您的应用程序。

df = structure(list(A = structure(c(1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L), .Label = c("x", "y"), class = "factor"), B = structure(c(3L, 3L, 3L, 2L, 2L, 2L, 1L, 1L, 1L, 1L), .Label = c("u", "w", "z"), class = "factor"), C = structure(c(1L, 2L, 1L, 1L, 2L, 1L, 2L, 2L, 1L, 1L), .Label = c("a1", "a2"), class = "factor")), .Names = c("A", "B", "C"), class = "data.frame", row.names = c("1", "2", "3", "4", "5", "6", "7", "8", "9", "10"))
library(data.table)
mingroup=2
setDT(df)[,n:=.N,.(A,B,C)][n<mingroup,c('A','B','C'):='other']

输出:

      A     B     C n
 1:     x     z    a1 2
 2: other other other 1
 3:     x     z    a1 2
 4: other other other 1
 5: other other other 1
 6: other other other 1
 7:     y     u    a2 2
 8:     y     u    a2 2
 9:     y     u    a1 2
10:     y     u    a1 2

---

备选方案:

df = structure(list(A = structure(c(1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L), .Label = c("x", "y"), class = "factor"), B = structure(c(3L, 3L, 3L, 2L, 2L, 2L, 1L, 1L, 1L, 1L), .Label = c("u", "w", "z"), class = "factor"), C = structure(c(1L, 2L, 1L, 1L, 2L, 1L, 2L, 2L, 1L, 1L), .Label = c("a1", "a2"), class = "factor")), .Names = c("A", "B", "C"), class = "data.frame", row.names = c("1", "2", "3", "4", "5", "6", "7", "8", "9", "10"))
df=setDT(df)
library(data.table)
mingroup=2
for(i in c('C','B','A'))
  df[,n:=.N,.(A,B,C)][n<mingroup,eval(i):='other'][,n:=NULL]

输出:

        A     B     C
 1:     x     z    a1
 2: other other other
 3:     x     z    a1
 4:     x     w other
 5:     x     w other
 6: other other other
 7:     y     u    a2
 8:     y     u    a2
 9:     y     u    a1
10:     y     u    a1