如果 column1 name == first(value) from column2 BY GROUP，则有条件地从 column1 取值

Question

我有这个假的 dataframe：

df <- structure(list(Group = c(1L, 1L, 2L, 2L), A = 1:4, B = 5:8, C = 9:12, 
X = c("A", "A", "B", "B")), class = "data.frame", row.names = c(NA, -4L))

  Group A B  C X
1     1 1 5  9 A
2     1 2 6 10 A
3     2 3 7 11 B
4     2 4 8 12 B

我尝试改变一个新列，它应该采用在另一列中具有列名的列的值：

所需的 output：

Group   A   B   C   X new_col
1       1   5   9     A 1
1       2   6   10    A 1
2       3   7   11    B 7
2       4   8   12    B 7

到目前为止我的尝试：

library(dplyr)

df %>% 
  group_by(Group) %>% 
  mutate(across(c(A,B,C), ~ifelse(first(X) %in% colnames(.), first(.), .), .names = "new_{.col}"))

  Group     A     B     C X     new_A new_B new_C
  <int> <int> <int> <int> <chr> <int> <int> <int>
1     1     1     5     9 A         1     5     9
2     1     2     6    10 A         1     5     9
3     2     3     7    11 B         3     7    11
4     2     4     8    12 B         3     7    11

Answer 1

一种选择可能是：

df %>%
    rowwise() %>%
    mutate(new_col = get(X)) %>%
    group_by(Group, X) %>%
    mutate(new_col = first(new_col))

 Group     A     B     C X     new_col
  <int> <int> <int> <int> <chr>   <int>
1     1     1     5     9 A           1
2     1     2     6    10 A           1
3     2     3     7    11 B           7
4     2     4     8    12 B           7

Answer 2

使用by并将+ 1添加到组号到 select 列。 假设组列按示例排列在"Group"列之后。

transform(df, new_col=do.call(rbind, by(df, df$Group, \(x) 
                                        cbind(paste(x$X, x[1, x$Group[1] + 1])))))
#   Group A B  C X new_col
# 1     1 1 5  9 A     A 1
# 2     1 2 6 10 A     A 1
# 3     2 3 7 11 B     B 7
# 4     2 4 8 12 B     B 7
          

注意： R version 4.1.2 (2021-11-01) 。

---

数据：

df <- structure(list(Group = c(1L, 1L, 2L, 2L), A = 1:4, B = 5:8, C = 9:12, 
    X = c("A", "A", "B", "B")), class = "data.frame", row.names = c(NA, 
-4L))

Answer 3

在base R中，我们可以使用行/列索引

df$new_col <- df[2:4][cbind(match(unique(df$Group), df$Group)[df$Group], 
       match(df$X, names(df)[2:4]))]
df$new_col
[1] 1 1 7 7