I'm attempting to use the collapsibleTree htmlwidget which requires a dataframe containing a row for each node, i.e. containing all parent-child relationships (known as a DataFrameNetwork in data.tree parlance I believe) in order to take advantage of some of the more useful features.
I am however starting with a dataframe containing a row for each leaf (known as a DataFrameTable?) and am struggling to get it in the right format. I believe the easiest thing to do is to use the data.tree package, but I'm open to suggestions, particularly if they're tidyverse orientated and avoid looping.
Example:
data:
lev1 <- c("A","A","A","B","B","C")
lev2 = c("a", letters[1:5])
lev3 = 1:6
df <- data.frame(lev1, lev2, lev3)
target:
result <- data.frame(parent = c(lev1, lev2),
child = c(lev2, lev3)) %>% unique()
(Awful code I know). This particular example has three levels in the hierarchy, but I need the solution to be general enough to apply to any number of levels. I've tried adding a pathString to the dataframe as per the documentation and following up with FromDataFrameTable() and then ToDataFrameNetwork() but I get the feeling I'm going about things all wrong.
library(zoo)
library(data.table)
library(dplyr)
rbindlist(lapply(as.data.frame(rollapply(names(df), 2, c), stringsAsFactors = F),
function(x) select(df, c(x)))) %>%
distinct() %>%
`colnames<-`(c("parent", "child"))
Output is:
parent child
1: A a
2: A b
3: B c
4: B d
5: C e
6: a 1
7: a 2
8: b 3
9: c 4
10: d 5
11: e 6
Sample data:
df <- structure(list(lev1 = structure(c(1L, 1L, 1L, 2L, 2L, 3L), .Label = c("A",
"B", "C"), class = "factor"), lev2 = structure(c(1L, 1L, 2L,
3L, 4L, 5L), .Label = c("a", "b", "c", "d", "e"), class = "factor"),
lev3 = 1:6), .Names = c("lev1", "lev2", "lev3"), row.names = c(NA,
-6L), class = "data.frame")