如何重塑具有多个级别的数据框

Question

我目前有一个格式如下所示的数据框（df1）：

ID	A1	A1.1	A1.2	A1.3	B1	B1.1	B1.2	B1.3	C1	C1.1	C1.2	C1.3
1	10	1	1	0	15	0	1	0	30	1	0	0
2	25	1	0	0	30	0	1	1	25	1	0	1
3	40	0	1	0	15	0	1	0	10	0	0	1
4	25	1	1	0	10	0	1	1	30	1	0	0

我想重新格式化它，使其按照 df2 中所示的方式排列：

ID	A1_C1_value	R1	R2	R3	A_C
1	10	1	1	0	一种
1	15	0	1	0	乙
1	30	1	0	0	C
2	25	1	0	0	一种
2	30	0	1	1	乙
2	25	1	0	1	C
3	40	0	1	0	一种
3	15	0	1	0	乙
3	10	0	0	1	C
4	25	1	1	0	一种
4	10	0	1	1	乙
4	30	1	0	0	C

我已经能够旋转更长的时间来获得 A1_C1 值和 A_C 列，但是剩下的是 A1.2-1.3、B1.2-1.3、C1.2-1.3。 我如何将这些包含在我的转换中？ 这是我到目前为止所做的：

df2 <- pivot_longer(df1, cols = c("A1","B1","C1"), names_to ="A_C",
 values_to = "A1_C1_value")

Answer 1

您可以使用pivot_longer()但如果您首先按如下方式重命名变量会更容易：

x <- data.frame(
  ID = 1:4, 
  A1 = c(10,25,40,25), 
  A1.1=c(1,1,0,1), 
  A1.2=c(1,0,1,1), 
  A1.3=c(0,0,0,0), 
  B1 = c(15,30,15,10), 
  B1.1=c(0,0,0,0), 
  B1.2=c(1,1,1,1), 
  B1.3=c(0,1,0,1), 
  C1 = c(30,25,10,30), 
  C1.1=c(1,1,0,1), 
  C1.2=c(0,0,0,0), 
  C1.3=c(0,1,1,0)
)

x %>% 
  rename("A1.0" = "A1", 
         "B1.0" = "B1", 
         "C1.0" = "C1") %>% 
  pivot_longer(`A1.0`:`C1.3`, 
               names_pattern=c("([A-C])\\d.(\\d)"), 
               names_to=c("A_C", ".value"), 
               names_prefix = "R") %>%
  rename("A1_C1_value" = "0", 
         "R1" = "1", 
         "R2" = "2", 
         "R3" = "3")

# # A tibble: 12 × 6
#      ID A_C   A1_C1_value    R1    R2    R3
#   <int> <chr>       <dbl> <dbl> <dbl> <dbl>
# 1     1 A              10     1     1     0
# 2     1 B              15     0     1     0
# 3     1 C              30     1     0     0
# 4     2 A              25     1     0     0
# 5     2 B              30     0     1     1
# 6     2 C              25     1     0     1
# 7     3 A              40     0     1     0
# 8     3 B              15     0     1     0
# 9     3 C              10     0     0     1
# 10     4 A              25     1     1     0
# 11     4 B              10     0     1     1
# 12     4 C              30     1     0     0**

Answer 2

您可以使用data.table非常有效地执行此data.table ：

library(data.table)
df1 <- data.table(df1)
df2 <- melt(df1, measure = patterns("[A-Z]1$", "1.1$", "1.2$", "1.3$"),
           value.name = c("A1_C1_value", "R1", "R2", "R3"), variable.name = "A_C")
df2[, A_C := LETTERS[A_C]][order(ID)]

生产：

    ID A_C A1_C1_value R1 R2 R3
 1:  1   A          10  1  1  0
 2:  1   B          15  0  1  0
 3:  1   C          30  1  0  0
 4:  2   A          25  1  0  0
 5:  2   B          30  0  1  1
 6:  2   C          25  1  0  1
 7:  3   A          40  0  1  0
 8:  3   B          15  0  1  0
 9:  3   C          10  0  0  1
10:  4   A          25  1  1  0
11:  4   B          10  0  1  1
12:  4   C          30  1  0  0