partykit (version 1.2-13)

party: Recursive Partytioning

Description

A class for representing decision trees and corresponding accessor functions.

Usage

party(node, data, fitted = NULL, terms = NULL, names = NULL, 
    info = NULL)
# S3 method for party
names(x)
# S3 method for party
names(x) <- value
data_party(party, id = 1L)
# S3 method for default
data_party(party, id = 1L)
node_party(party)
is.constparty(party)
is.simpleparty(party)

Arguments

node

an object of class partynode.

data

a (potentially empty) data.frame.

fitted

an optional data.frame with nrow(data) rows (only if nrow(data) != 0 and containing at least the fitted terminal node identifiers as element (fitted). In addition, weights may be contained as element (weights) and responses as (response).

terms

an optional terms object.

names

an optional vector of names to be assigned to each node of node.

info

additional information.

x

an object of class party.

party

an object of class party.

value

a character vector of up to the same length as x, or NULL.

id

a node identifier.

Value

The constructor returns an object of class party:

node

an object of class partynode.

data

a (potentially empty) data.frame.

fitted

an optional data.frame with nrow(data) rows (only if nrow(data) != 0 and containing at least the fitted terminal node identifiers as element (fitted). In addition, weights may be contained as element (weights) and responses as (response).

terms

an optional terms object.

names

an optional vector of names to be assigned to each node of node.

info

additional information.

names can be used to set and retrieve names of nodes and node_party returns an object of class partynode. data_party returns a data frame with observations contained in node id.

Details

Objects of class party basically consist of a partynode object representing the tree structure in a recursive way and data. The data argument takes a data.frame which, however, might have zero columns. Optionally, a data.frame with at least one variable (fitted) containing the terminal node numbers of data used for fitting the tree may be specified along with a terms object or any additional (currently unstructured) information as info. Argument names defines names for all nodes in node.

Method names can be used to extract or alter names for nodes. Function node_party returns the node element of a party object. Further methods for party objects are documented in party-methods and party-predict. Trees of various flavors can be coerced to party, see party-coercion.

Two classes inherit from class party and impose additional assumptions on the structure of this object: Class constparty requires that the fitted slot contains a partitioning of the learning sample as a factor ("fitted") and the response values of all observations in the learning sample as ("response"). This structure is most flexible and allows for graphical display of the response values in terminal nodes as well as for computing predictions based on arbitrary summary statistics.

Class simpleparty assumes that certain pre-computed information about the distribution of the response variable is contained in the info slot nodes. At the moment, no formal class is used to describe this information.

References

Hothorn T, Zeileis A (2015). partykit: A Modular Toolkit for Recursive Partytioning in R. Journal of Machine Learning Research, 16, 3905--3909.

Examples

Run this code
# NOT RUN {
### data ###
## artificial WeatherPlay data
data("WeatherPlay", package = "partykit")
str(WeatherPlay)


### splits ###
## split in overcast, humidity, and windy
sp_o <- partysplit(1L, index = 1:3)
sp_h <- partysplit(3L, breaks = 75)
sp_w <- partysplit(4L, index = 1:2)

## query labels
character_split(sp_o)


### nodes ###
## set up partynode structure
pn <- partynode(1L, split = sp_o, kids = list(
  partynode(2L, split = sp_h, kids = list(
    partynode(3L, info = "yes"),
    partynode(4L, info = "no"))),
  partynode(5L, info = "yes"),
  partynode(6L, split = sp_w, kids = list(
    partynode(7L, info = "yes"),
    partynode(8L, info = "no")))))
pn


### tree ###
## party: associate recursive partynode structure with data
py <- party(pn, WeatherPlay)
py
plot(py)


### variations ###
## tree stump
n1 <- partynode(id = 1L, split = sp_o, kids = lapply(2L:4L, partynode))
print(n1, data = WeatherPlay)

## query fitted nodes and kids ids
fitted_node(n1, data = WeatherPlay)
kidids_node(n1, data = WeatherPlay)

## tree with full data sets
t1 <- party(n1, data = WeatherPlay)

## tree with empty data set
party(n1, data = WeatherPlay[0, ])

## constant-fit tree
t2 <- party(n1, 
  data = WeatherPlay,
  fitted = data.frame(
    "(fitted)" = fitted_node(n1, data = WeatherPlay),
    "(response)" = WeatherPlay$play,
    check.names = FALSE),
  terms = terms(play ~ ., data = WeatherPlay),
)
t2 <- as.constparty(t2)
t2
plot(t2)
# }

Run the code above in your browser using DataCamp Workspace