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) Deprecated. Update to the other constructor that accepts the predictor indices in argument 10. Creates the root node of a decision tree and contains information about the relationship of child nodes.
Tree(int nRows, int minObsAllowedInChild, int maxNumberOfCategories, int maxNumberOfNodes, int nClasses, int nPreds, PredictiveModel.VariableType varType, int[] predNValues, PredictiveModel.VariableType[] predType, int[] predIdx, int nyMissing) Creates the root node of a decision tree and contains information about the relationship of child nodes.

Method Summary

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.
int[]	getNodeAssigments(double[][] testData) Returns the node assignments for testData using the 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.
double[]	getPredictions(double[][] testData) Returns the predicted values on testData using the tree.
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	printTree() Prints the tree structure.

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)

Deprecated. Update to the other constructor that accepts the predictor indices in argument 10. Creates the root node of a decision tree and contains information about the relationship of child nodes.

Parameters:

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

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

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

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

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

nPreds - an int, 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 containing the variable type of each predictor variable

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

Tree

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

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

Parameters:

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

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

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

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

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

nPreds - an int, 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 containing the variable type of each predictor variable

predIdx - an int array containing the column index of each predictor

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

Method Detail

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

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.

printTree

public void printTree()

Prints the tree structure.

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 int, 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, the number of nodes or size of the tree

getNumberOfPredictors

public int getNumberOfPredictors()

Returns the number of predictors used in the model.

Returns:

an int, 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 relevant only for classes that implement the com.imsl.datamining.decisionTree.DecisionTreeSurrogateMethod interface.

Returns:

an int, the number of surrogate splits searched for at each node

getPredictorNumberOfValues

public int[] getPredictorNumberOfValues()

Returns the number of distinct values of each predictor variable.

For continuous predictor variables, the value is set to 0 and is not meaningful.

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, the index of the predictor

Returns:

a VariableType, 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, indicates the response variable type

isTerminalNode

public boolean isTerminalNode(int i)

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

Parameters:

i - an int specifying the index of the node

Returns:

a boolean, that if true, then the node at index i is a terminal node

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

getNodeAssigments

public int[] getNodeAssigments(double[][] testData)

Returns the node assignments for testData using the tree.

The user must ensure that testData has the same number of columns, variable types, and variable column locations as the original training data.

Parameters:

testData - a double matrix

Returns:

an int array containing the node assignments for each row of the input matrix

getPredictions

public double[] getPredictions(double[][] testData)

Returns the predicted values on testData using the tree.

The user must ensure that testData has the same number of columns, variable types, and variable column locations as the original training data.

Parameters:

testData - a double matrix

Returns:

an int array containing the node assignments for each row of the input matrix