在两个变量中将wide转换为long

Question

我有这样的数据帧

df= data.frame(a1 = c(1,2,3), a2 = c(4,5,6), b1 = c(1,2,3), b2= c(4,NaN,6), id = c(1,2,3))

我想得到

  id  a measure1 b  measure2
1  1 a1        1 b1        1
2  2 a1        2 b1        2
3  3 a1        3 b1        3
4  1 a2        4 b2        4
5  2 a2        5 b2      NaN
6  3 a2        6 b2        6

我会做

df1 = df[, c(1,2,5)]
df2 = df[, c(3,4,5)]
library(reshape2)
df1_long = melt(df1,id.vars= 'id', measure.vars=c("a1", "a2"),
                               variable.name="a",
                value.name="measure1")
df2_long = melt(df2,id.vars= 'id', measure.vars=c("b1", "b2"),
                variable.name="b",
                value.name="measure2")

df_new = cbind(df1_long, df2_long[, -1])

但我认为有一种更简单的方法

Answer 1

在基数R中，您可以使用reshape功能。 然后，您可以根据需要更改列：

transform(reshape(df,1:4,sep="",dir="long"),a_=paste0("a",time),b_=paste0("b",time))

    id time a   b a_ b_
1.1  1    1 1   1 a1 b1
2.1  2    1 2   2 a1 b1
3.1  3    1 3   3 a1 b1
1.2  1    2 4   4 a2 b2
2.2  2    2 5 NaN a2 b2
3.2  3    2 6   6 a2 b2

Answer 2

这是你认为更容易的相对，但一个选项是使用dplyr和tidyr ：

df %>%
 select(id, starts_with("a")) %>%
 gather(a, measurement1, -id) %>%
 bind_cols(df %>%
            select(starts_with("b")) %>%
            gather(b, measurement2))

  id  a measurement1  b measurement2
1  1 a1            1 b1            1
2  2 a1            2 b1            2
3  3 a1            3 b1            3
4  1 a2            4 b2            4
5  2 a2            5 b2          NaN
6  3 a2            6 b2            6