smile.base.cart

Class InternalNode

java.lang.Object

smile.base.cart.InternalNode

All Implemented Interfaces:

java.io.Serializable, Node

Direct Known Subclasses:

NominalNode, OrdinalNode

public abstract class InternalNode
extends java.lang.Object
implements Node

An internal node in CART.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
InternalNode(int feature, double score, double deviance, Node trueChild, Node falseChild)

Method Summary

Modifier and Type	Method and Description
abstract boolean	branch(smile.data.Tuple x) Returns true if the instance goes to the true branch.
int	depth() Returns the maximum depth of the tree -- the number of nodes along the longest path from this node down to the farthest leaf node.
double	deviance() Returns the deviance of node.
Node	falseChild() Returns the false branch child.
int	feature() Returns the split feature.
int	leafs() Returns the number of leaf nodes in the subtree.
Node	merge() Try to merge the children nodes and return a leaf node.
abstract LeafNode	predict(smile.data.Tuple x) Evaluates the tree over an instance.
abstract InternalNode	replace(Node trueChild, Node falseChild) Returns a new internal node with children replaced.
double	score() Returns the split score (reduction of impurity).
int	size() Returns the number of samples in the node.
abstract java.lang.String	toString(smile.data.type.StructType schema, boolean trueBranch) Returns the string representation of branch.
int[]	toString(smile.data.type.StructType schema, smile.data.type.StructField response, InternalNode parent, int depth, java.math.BigInteger id, java.util.List<java.lang.String> lines) Adds the string representation (R's rpart format) to a collection.
Node	trueChild() Returns the true branch child.

Methods inherited from class java.lang.Object

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

Methods inherited from interface smile.base.cart.Node

dot

Constructor Detail

InternalNode

public InternalNode(int feature,
                    double score,
                    double deviance,
                    Node trueChild,
                    Node falseChild)

Method Detail

predict

public abstract LeafNode predict(smile.data.Tuple x)

Evaluates the tree over an instance.

Specified by:

predict in interface Node

branch

public abstract boolean branch(smile.data.Tuple x)

Returns true if the instance goes to the true branch.

replace

public abstract InternalNode replace(Node trueChild,
                                     Node falseChild)

Returns a new internal node with children replaced.

trueChild

public Node trueChild()

Returns the true branch child.

falseChild

public Node falseChild()

Returns the false branch child.

feature

public int feature()

Returns the split feature.

score

public double score()

Returns the split score (reduction of impurity).

size

public int size()

Description copied from interface: Node

Returns the number of samples in the node.

Specified by:

size in interface Node

leafs

public int leafs()

Description copied from interface: Node

Returns the number of leaf nodes in the subtree.

Specified by:

leafs in interface Node

deviance

public double deviance()

Description copied from interface: Node

Returns the deviance of node.

Specified by:

deviance in interface Node

depth

public int depth()

Description copied from interface: Node

Returns the maximum depth of the tree -- the number of nodes along the longest path from this node down to the farthest leaf node.

Specified by:

depth in interface Node

merge

public Node merge()

Description copied from interface: Node

Try to merge the children nodes and return a leaf node. If not able to merge, return this node itself.

Specified by:

merge in interface Node

toString

public abstract java.lang.String toString(smile.data.type.StructType schema,
                                          boolean trueBranch)

Returns the string representation of branch.

Parameters:

schema - the schema of data.

trueBranch - for true or false branch.

toString

public int[] toString(smile.data.type.StructType schema,
                      smile.data.type.StructField response,
                      InternalNode parent,
                      int depth,
                      java.math.BigInteger id,
                      java.util.List<java.lang.String> lines)

Description copied from interface: Node

Adds the string representation (R's rpart format) to a collection.

Specified by:

toString in interface Node

Parameters:

schema - the schema of data

response - the schema of response variable

parent - the parent node

depth - the depth of node in the tree. The root node is at depth 0.

id - node id

lines - the collection of node's string representation.

Returns:

the sample count of each class for decision tree; single element array [node size] for regression tree.