Learn R Programming

partykit (version 1.2-23)

party-methods: Methods for Party Objects

Description

Methods for computing on party objects.

Usage

# S3 method for party
print(x, 
    terminal_panel = function(node)
      formatinfo_node(node, default = "*", prefix = ": "),
    tp_args = list(),
    inner_panel = function(node) "", ip_args = list(),
    header_panel = function(party) "",
    footer_panel = function(party) "",
    digits = getOption("digits") - 2, ...)
# S3 method for simpleparty
print(x, digits = getOption("digits") - 4,
    header = NULL, footer = TRUE, ...)
# S3 method for constparty
print(x, FUN = NULL, digits = getOption("digits") - 4,
    header = NULL, footer = TRUE, ...)
# S3 method for party
length(x)
# S3 method for party
[(x, i, ...)
# S3 method for party
[[(x, i, ...)
# S3 method for party
depth(x, root = FALSE, ...)
# S3 method for party
width(x, ...)
# S3 method for party
nodeprune(x, ids, ...)

Arguments

x

an object of class party.

i

an integer specifying the root of the subtree to extract.

terminal_panel

a panel function for printing terminal nodes.

tp_args

a list containing arguments to terminal_panel.

inner_panel

a panel function for printing inner nodes.

ip_args

a list containing arguments to inner_panel.

header_panel

a panel function for printing the header.

footer_panel

a panel function for printing the footer.

digits

number of digits to be printed.

header

header to be printed.

footer

footer to be printed.

FUN

a function to be applied to nodes.

root

a logical. Should the root count be counted in depth?

ids

a vector of node ids (or their names) to be pruned-off.

...

additional arguments.

Details

length gives the number of nodes in the tree (in contrast to the length method for partynode objects which returns the number of kid nodes in the root), depth the depth of the tree and width the number of terminal nodes. The subset methods extract subtrees and the print method generates a textual representation of the tree. nodeprune prunes-off nodes and makes sure that the node ids of the resulting tree are in pre-order starting with root node id 1. For constparty objects, the fitted slot is also changed.

Examples

Run this code

    ## a tree as flat list structure
    nodelist <- list(
        # root node
        list(id = 1L, split = partysplit(varid = 4L, breaks = 1.9), 
            kids = 2:3),
        # V4 <= 1.9, terminal node
        list(id = 2L),
        # V4 > 1.9
        list(id = 3L, split = partysplit(varid = 5L, breaks = 1.7), 
            kids = c(4L, 7L)),
        # V5 <= 1.7
        list(id = 4L, split = partysplit(varid = 4L, breaks = 4.8), 
            kids = 5:6),
        # V4 <= 4.8, terminal node
        list(id = 5L),
        # V4 > 4.8, terminal node
        list(id = 6L),
        # V5 > 1.7, terminal node
        list(id = 7L)
    )

    ## convert to a recursive structure
    node <- as.partynode(nodelist)

    ## set up party object
    data("iris")
    tree <- party(node, data = iris, 
        fitted = data.frame("(fitted)" = 
            fitted_node(node, data = iris),
            check.names = FALSE))
    names(tree) <- paste("Node", nodeids(tree), sep = " ")

    ## number of kids in root node
    length(tree)

    ## depth of tree
    depth(tree)

    ## number of terminal nodes
    width(tree)

    ## node number four
    tree["Node 4"]
    tree[["Node 4"]]

Run the code above in your browser using DataLab