com.imsl.datamining.decisionTree

Class Tree

java.lang.Object

com.imsl.datamining.decisionTree.Tree

All Implemented Interfaces:

Serializable, Cloneable

public class Tree
extends Object
implements Serializable, Cloneable

Serves as the root node of a decision tree and contains information about the relationship of child nodes.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
Tree(int nRows, int minObsAllowedInChild, int maxNumberOfCategories, int maxNumberOfNodes, int nClasses, int nPreds, PredictiveModel.VariableType varType, int[] predNValues, PredictiveModel.VariableType[] predType, int nyMissing) Creates the root node of a decision tree and contains information about the relationship of child nodes.

Method Summary

Methods

Modifier and Type	Method and Description
protected Tree	clone() Returns a clone of this object.
TreeNode	getNode(int i) Returns a copy of the specified node of the decision tree.
TreeNode[]	getNodes() Returns nodes within a decision tree.
int	getNumberOfClasses() Returns the number of classes assumed by the response variable, if the response variable is categorical.
int	getNumberOfLevels() Returns the number of levels or depth of a tree.
int	getNumberOfNodes() Returns the number of nodes (size of a tree).
int	getNumberOfPredictors() Returns the number of predictors used in the model.
int	getNumberOfSurrogateSplits() Returns the number of surrogate splits searched for at each tree node.
int[]	getPredictorNumberOfValues() Returns the number of distinct values of each predictor variable.
PredictiveModel.VariableType	getPredictorType(int i) Returns the PredictiveModel.VariableType of a predictor variable.
PredictiveModel.VariableType	getResponseType() Returns the PredictiveModel.VariableType of the response variable.
boolean[]	getTerminalNodeIndicators() Returns the terminal node indicator array.
boolean	isTerminalNode(int i) Returns the terminal node indicator of the node at the given index.
void	setNode(int i, TreeNode dtn) Inserts a TreeNode into the specified location of the decision tree.
void	setNodes(TreeNode[] nodes) Sets the nodes within a decision tree.
void	setNumberOfLevels(int nLevels) Sets the number of levels determined for a tree (depth).
void	setNumberOfNodes(int nNodes) Sets the number of nodes (size of a tree).
void	setNumberOfSurrogateSplits(int nSurrogates) Sets the number of surrogate splits to search for at each tree node.
void	setTerminalNode(int i, boolean terminalNode) Sets the terminal node indicator of the node at the given index.
void	setTerminalNodeIndicators(boolean[] terminalNodes) Sets the terminal node indicator array.

Methods inherited from class java.lang.Object

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

Constructor Detail

Tree

public Tree(int nRows,
    int minObsAllowedInChild,
    int maxNumberOfCategories,
    int maxNumberOfNodes,
    int nClasses,
    int nPreds,
    PredictiveModel.VariableType varType,
    int[] predNValues,
    PredictiveModel.VariableType[] predType,
    int nyMissing)

Creates the root node of a decision tree and contains information about the relationship of child nodes.

Parameters:

nRows - an int indicating the number of observations (rows) in the input data.

minObsAllowedInChild - an int indicating the number of observations necessary in all potential child nodes before a node may be split.

maxNumberOfCategories - an int indicating the maximum number of categories allowed for categorical predictor variables.

maxNumberOfNodes - an int indicating the maximum number of nodes allowed in the tree.

nClasses - an int indicating the number of classes assumed by the response variable, if the response variable is categorical.

nPreds - an int indicating the number of predictors used in the model.

varType - a PredictiveModel.VariableType array of length equal to the number of variables active in the model, which is less than or equal to the available columns in the data, (xy[i].length).

predNValues - an int array containing the number of values of each predictor variable.

predType - a PredictiveModel.VariableType array of length equal to the number of predictors containing the type of each predictor variable.

nyMissing - an int equal to the number of missing values in the response variable.

Method Detail

clone

protected Tree clone()

Returns a clone of this object.

Overrides:

clone in class Object

Returns:

a Tree that is a copy of this object.

getNode

public TreeNode getNode(int i)

Returns a copy of the specified node of the decision tree.

Parameters:

i - an int that specifies the index of a TreeNode to be returned.

Returns:

a TreeNode corresponding to the specified index.

getNodes

public TreeNode[] getNodes()

Returns nodes within a decision tree.

Returns:

a TreeNode array containing a copy of the TreeNodes within the decision tree.

getNumberOfClasses

public int getNumberOfClasses()

Returns the number of classes assumed by the response variable, if the response variable is categorical.

Returns:

an intequal to the number of classes assumed by the response variable, if the response variable is categorical.

getNumberOfLevels

public int getNumberOfLevels()

Returns the number of levels or depth of a tree.

Returns:

an int specifying the number of levels or depth of a tree.

getNumberOfNodes

public int getNumberOfNodes()

Returns the number of nodes (size of a tree).

Returns:

an int specifying the number of nodes in a tree (size).

getNumberOfPredictors

public int getNumberOfPredictors()

Returns the number of predictors used in the model.

Returns:

an int equal to the number of predictors used in the model.

getNumberOfSurrogateSplits

public int getNumberOfSurrogateSplits()

Returns the number of surrogate splits searched for at each tree node.

Surrogate splits are only relevant for classes which implement the com.imsl.datamining.decisionTree.DecisionTreeSurrogateMethod interface.

Returns:

an int equal to the number of surrogate splits searched for at each node.

getPredictorNumberOfValues

public int[] getPredictorNumberOfValues()

Returns the number of distinct values of each predictor variable.

Note that if the predictor is com.imsl.datamining.PredictiveModel.VariableType.QUANTITATIVE_CONTINUOUS, the corresponding value in the array is set to 0.

Returns:

an int array containing the number of values of each predictor variable.

getPredictorType

public PredictiveModel.VariableType getPredictorType(int i)

Returns the PredictiveModel.VariableType of a predictor variable.

Parameters:

i - an int specifying the predictor.

Returns:

a VariableType equal to the variable type of the predictor at index i.

getResponseType

public PredictiveModel.VariableType getResponseType()

Returns the PredictiveModel.VariableType of the response variable.

Returns:

a PredictiveModel.VariableType that indicates the response variable type.

getTerminalNodeIndicators

public boolean[] getTerminalNodeIndicators()

Returns the terminal node indicator array.

Returns:

a boolean array indicating which nodes are terminal nodes and which are not. If true at index i, then the node at index i is a terminal node.

isTerminalNode

public boolean isTerminalNode(int i)

Returns the terminal node indicator of the node at the given index.

Parameters:

i - an int which indicates what node to query.

Returns:

a boolean indicating whether the specified node is a terminal node or not.

setNode

public void setNode(int i,
           TreeNode dtn)

Inserts a TreeNode into the specified location of the decision tree.

Parameters:

i - an int indicating the index for insertion of a TreeNode.

dtn - a TreeNode to be inserted into the decision tree.

setNodes

public void setNodes(TreeNode[] nodes)

Sets the nodes within a decision tree.

Parameters:

nodes - a TreeNode array specifying TreeNodes within the decision tree.

setNumberOfLevels

public void setNumberOfLevels(int nLevels)

Sets the number of levels determined for a tree (depth).

Parameters:

nLevels - an int indicating the number of levels determined for a tree (depth).

nLevels must be nonnegative.

Default: nLevels = 0

setNumberOfNodes

public void setNumberOfNodes(int nNodes)

Sets the number of nodes (size of a tree).

Parameters:

nNodes - an int specifying the number of nodes in a tree (size).

nNodes must be nonnegative.

Default: nNodes = 0

setNumberOfSurrogateSplits

public void setNumberOfSurrogateSplits(int nSurrogates)

Sets the number of surrogate splits to search for at each tree node.

Surrogate splits are only relevant for classes which implement the com.imsl.datamining.decisionTree.DecisionTreeSurrogateMethod interface.

Parameters:

nSurrogates - an int specifying the number of surrogate splits to search for at each tree node.

nSurrogates must be nonnegative and less than the number of predictors - 1.

Default: nSurrogates = 0

setTerminalNode

public void setTerminalNode(int i,
                   boolean terminalNode)

Sets the terminal node indicator of the node at the given index.

Parameters:

i - an int which indicates which node.

terminalNode - a boolean indicating whether the specified node is a terminal node or not (true for a terminal node false otherwise).

Default: terminalNode = false.

setTerminalNodeIndicators

public void setTerminalNodeIndicators(boolean[] terminalNodes)

Sets the terminal node indicator array.

Parameters:

terminalNodes - a boolean array of terminal node designations. If true at index i, then the node at index i is a terminal node.

Default: terminalNodes[i] = false.