R: Pairing of consecutive items into new separate columns

Question

Given a table with a column of non-unique items (here, Name) and two more columns: one with labels (Group) that are unique only within the group of the first columns, and a third containing values (Delay), I'm trying to end up with a table containing columns for Name and Group, and a third column with the Group of the next-highest Delay value. If there is no next-highest, then there is no row.

So for a dataset such as:

df = data.frame(
  Name =    c('lorem', 'lorem', 'lorem', 'lorem', 'lorem', 'ipsum', 'ipsum', 'ipsum', 'ipsum', 'ipsum', 'dolor', 'dolor', 'dolor', 'dolor', 'dolor', 'dolor', 'amet', 'amet', 'amet', 'amet', 'amet', 'amet'), 
  Group =   c('E', 'D', 'C', 'B', 'A', 'E', 'D', 'C', 'B', 'A', 'C', 'A', 'B', 'D', 'F', 'E', 'C', 'A', 'B', 'D', 'F', 'E'), 
  Delay =   c(5, 32, 59, 86, 113, 0, 27, 54, 81, 108, 10, 37, 64, 91, 111, 118, 0, 27, 54, 81, 101, 108)
  )

   Name Group Delay
1  lorem     E     5
2  lorem     D    32
3  lorem     C    59
4  lorem     B    86
5  lorem     A   113
6  ipsum     E     0
7  ipsum     D    27
8  ipsum     C    54
9  ipsum     B    81
10 ipsum     A   108
11 dolor     C    10
12 dolor     A    37
13 dolor     B    64
14 dolor     D    91
15 dolor     F   111
16 dolor     E   118
17  amet     C     0
18  amet     A    27
19  amet     B    54
20  amet     D    81
21  amet     F   101
22  amet     E   108

The desired output would be (although keeping the higher Delay value for each pair wouldn't hurt):

Name    Source  Target
lorem   E       D
lorem   D       C
lorem   C       B
ipsum   B       A
ipsum   E       D
ipsum   D       C
ipsum   C       B
ipsum   B       A
dolor   C       A
dolor   A       B
dolor   B       D
dolor   D       F
dolor   F       E
amet    C       A
amet    A       B
amet    B       D
amet    F       E

Ultimately, this will go into a sankeyNetwork graph using the networkD3 package.

I did try the following, looking at the next row for a match in the Name (after sorting), although this didn't work as expected on my actual data but also does nothing on the dummy data:

l = data.frame(Name = character(), From = character(), Target = character())

for(i in 1:(nrow(df) - 1)){
  if(df$Name[i] == df$Name[i + 1])
  {
    From = as.character(df$Group[i])
    Target = as.character(df$Group[i + 1])
    Name = as.character(df$Name[i])
  }
  links = rbind(l, list(Name = as.character(Name), From = as.character(From), Target = as.character(Target)))
}

Answer 1

We can do a group by and take the lead

library(dplyr)
df %>%
   group_by(Name) %>%
   transmute(Source = Group, 
             Target = lead(Group, order_by = Delay),
             Value = lead(Delay, order_by = Delay)) %>%
   na.omit