marytts.cart

Class CART

java.lang.Object

marytts.cart.DirectedGraph

marytts.cart.CART

Direct Known Subclasses:

FeatureVectorCART, StringPredictionTree

public class CART
extends DirectedGraph

A tree is a specific kind of directed graph in which each node can have only a single parent node. It consists exclusively of DecisionNode and LeafNode nodes.

Author:

marc

Field Summary

Fields inherited from class marytts.cart.DirectedGraph

featDef, logger, properties, rootNode

Constructor Summary

Constructors

Constructor and Description
CART() Build a new empty cart
CART(FeatureDefinition featDef) Build a new empty cart with the given feature definition.
CART(Node rootNode, FeatureDefinition featDef) Build a new cart with the given node as the root node
CART(Node rootNode, FeatureDefinition featDef, Properties properties) Build a new cart with the given node as the root node

Method Summary

Methods

Modifier and Type	Method and Description
Object	interpret(Target target, int minNumberOfData) Passes the given item through this CART and returns the interpretation.
Node	interpretToNode(FeatureVector featureVector, int minNumberOfData) Passes the given item through this CART and returns the leaf Node, or the Node it stopped walking down.
Node	interpretToNode(Target target, int minNumberOfData) Passes the given item through this CART and returns the leaf Node, or the Node it stopped walking down.
static Node	replaceLeafByCart(CART cart, LeafNode leaf) In this tree, replace the given leaf with the given CART

Methods inherited from class marytts.cart.DirectedGraph

getDecisionNodeIterator, getDecisionNodes, getDirectedGraphNodeIterator, getDirectedGraphNodes, getFeatureDefinition, getLeafNodeIterator, getLeafNodes, getNodeIterator, getNodes, getNumNodes, getProperties, getRootNode, interpret, interpret, interpret, setRootNode, toString

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

CART

public CART()

Build a new empty cart

CART

public CART(FeatureDefinition featDef)

Build a new empty cart with the given feature definition.

Parameters:

featDef - featDef

CART

public CART(Node rootNode,
    FeatureDefinition featDef)

Build a new cart with the given node as the root node

Parameters:

rootNode - the root node of the CART

featDef - the feature definition used for interpreting the meaning of decision node criteria.

CART

public CART(Node rootNode,
    FeatureDefinition featDef,
    Properties properties)

Build a new cart with the given node as the root node

Parameters:

rootNode - the root node of the CART

featDef - the feature definition used for interpreting the meaning of decision node criteria.

properties - a generic properties object, which can be used to encode information about the tree and the way the data in it should be represented.

Method Detail

interpretToNode

public Node interpretToNode(Target target,
                   int minNumberOfData)

Passes the given item through this CART and returns the leaf Node, or the Node it stopped walking down.

Parameters:

target - the target to analyze

minNumberOfData - the minimum number of data requested. If this is 0, walk down the CART until the leaf level.

Returns:

the Node

interpretToNode

public Node interpretToNode(FeatureVector featureVector,
                   int minNumberOfData)

Passes the given item through this CART and returns the leaf Node, or the Node it stopped walking down.

Parameters:

featureVector - the target to analyze

minNumberOfData - the minimum number of data requested. If this is 0, walk down the CART until the leaf level.

Returns:

the Node

interpret

public Object interpret(Target target,
               int minNumberOfData)

Passes the given item through this CART and returns the interpretation.

Parameters:

target - the target to analyze

minNumberOfData - the minimum number of data requested. If this is 0, walk down the CART until the leaf level.

Returns:

the interpretation

replaceLeafByCart

public static Node replaceLeafByCart(CART cart,
                     LeafNode leaf)

In this tree, replace the given leaf with the given CART

Parameters:

cart - the CART

leaf - the leaf

Returns:

the ex-root node from cart which now replaces leaf.