For a given partynode or respective party object, I want to add a new split.
library(partykit)
sp_o <- partysplit(1L, index = 1:3)
sp_h <- partysplit(3L, breaks = 75)
node <- partynode(1L, split = sp_o, kids = list(
partynode(2L, split = sp_h, kids = list(
partynode(3L),
partynode(4L))),
partynode(5L)))
print(node)
##[1] root
##| [2] V1 in (-Inf,1]
##| | [3] V3 <= 75 *
##| | [4] V3 > 75 *
##| [5] V1 in (1,2] *
Now I want to add a new split to terminal node, e.g. node 3. I found the following way to do this:
# New split
sp_w <- partysplit(4L, index = 1:2)
# Access terminal node 3 and add the new split
node$kids[[1]]$kids[[1]] <- partynode(3L, split = sp_w, kids = list(
partynode(4L),
partynode(5L)))
node <- as.partynode(node)
print(node)
##[1] root
##| [2] V1 in (-Inf,1]
##| | [3] V3 <= 75
##| | | [4] V4 <= 1 *
##| | | [5] V4 > 1 *
##| | [6] V3 > 75 *
##| [7] V1 in (1,2] *
Is there way to do this that avoids having to call $kids[[i]] repeatedly? Assume that I know the vector of indices to access the respective node (e.g., here: c(1,1)). I want to do this for partynodes of arbitrary complexity, simply using a vector of indices to refer to the position of the desired terminal node (e.g., c(1,1,2)), or just the terminal node id, if possible.
You can do this by converting between list objects and partynode objects:
Convert original partynode to list
li <- as.list(node)
Node id to replace and maximum node id
i <- 3L
n <- length(li)
Set up new partynode and convert to list
node2 <- partynode(i, split = sp_w, kids = list(partynode(n + 1L), partynode(n + 2L)))
li2 <- as.list(node2)
Replace node id with new root node and append the rest
li[[i]] <- li2[[1L]]
li <- c(li, li2[-1L])
Convert back to partynode
node <- as.partynode(li)
print(node)
## [1] root
## | [2] V1 in (-Inf,1]
## | | [3] V3 <= 75
## | | | [4] V4 <= 1 *
## | | | [5] V4 > 1 *
## | | [6] V3 > 75 *
## | [7] V1 in (1,2] *