使用 dplyr 创建共识列

Question

我有一个数据框：

  Groups    Name    Category    value
        G1  A   cat1    20
        G1  A   cat2    1
        G1  B   cat3    21
        G1  B   cat3    23
        G2  B   cat4    32
        G2  C   cat2    23
        G2  C   cat2    21

我想添加一个新列consensus_category，例如：

Groups  Name    Category    value   consensus_category
G1  A   cat1    20  cat2
G1  A   cat2    1   cat2
G1  B   cat3    21  cat2
G1  B   cat3    23  cat2
G2  A   cat4    32  cat4
G2  C   cat2    23  cat4
G2  C   cat2    21  cat4

这个想法是我有一个 vector = c("A") 对应于数据框中的特定名称

从这个名字，我想为同一Groups中的所有其他行写对应的Category，但如果在两个Categories之间有一个ex-aequo，那么获胜者就是@987654329 @ 最低的Value。 （如：

G1  A   cat1    20  cat2
G1  A   cat2    1   cat2

cat2 获胜，因为1 < 20

我试过了：

df %>%
  group_by(Groups) %>%
  add_count(Category) %>%
  top_n(1, n) %>%
  top_n(-1, Value) %>%
  distinct(consensus_category = Category) %>%
  right_join(df) 

但我不知道如何指定我想要作为共识指南向量(A) 中的值。

Answer 1

使用dplyr可以找到组内有vec的Name，得到最小的value并从中提取出对应的Category。这是假设每个Groups 都至少有一个vec 的值。

library(dplyr)

vec <- "A"

df %>%
  group_by(Groups) %>%
  mutate(consensus_category = Category[value == min(value[Name == vec])])

#  Groups Name  Category value consensus_category
#  <fct>  <fct> <fct>    <int> <fct>             
#1 G1     A     cat1        20 cat2              
#2 G1     A     cat2         1 cat2              
#3 G1     B     cat3        21 cat2              
#4 G1     B     cat3        23 cat2              
#5 G2     A     cat4        32 cat4              
#6 G2     C     cat2        23 cat4              
#7 G2     C     cat2        21 cat4      

如果vec 中有多个值，您可能需要Name %in% vec 而不是==。

数据

df <- structure(list(Groups = c("G1", "G1", "G1", "G1", "G2", "G2", 
"G2"), Name = c("A", "A", "B", "B", "A", "C", "C"), Category = 
c("cat1", "cat2", "cat3", "cat3", "cat4", "cat2", "cat2"), value = 
c(20L, 1L, 21L, 23L, 32L, 23L, 21L)), class = "data.frame", row.names = 
c(NA, -7L))

Answer 2

data.table 的选项

library(data.table)
setDT(df)[, consensus_category := Category[value ==
      min(value[Name == vec])],  Groups]
df
#   Groups Name Category value consensus_category
#1:     G1    A     cat1    20               cat2
#2:     G1    A     cat2     1               cat2
#3:     G1    B     cat3    21               cat2
#4:     G1    B     cat3    23               cat2
#5:     G2    A     cat4    32               cat4
#6:     G2    C     cat2    23               cat4
#7:     G2    C     cat2    21               cat4

数据

df <- structure(list(Groups = c("G1", "G1", "G1", "G1", "G2", "G2", 
"G2"), Name = c("A", "A", "B", "B", "A", "C", "C"), Category = 
c("cat1", "cat2", "cat3", "cat3", "cat4", "cat2", "cat2"), value = 
c(20L, 1L, 21L, 23L, 32L, 23L, 21L)), class = "data.frame", row.names = 
c(NA, -7L))