org.processmining.datadiscovery.estimators.impl

Class DecisionTreeFunctionEstimator

java.lang.Object

org.processmining.datadiscovery.estimators.AbstractFunctionEstimator

org.processmining.datadiscovery.estimators.AbstractDecisionTreeFunctionEstimator

org.processmining.datadiscovery.estimators.impl.DecisionTreeFunctionEstimator

All Implemented Interfaces:

org.processmining.datadiscovery.estimators.DecisionTreeBasedFunctionEstimator, org.processmining.datadiscovery.estimators.FunctionEstimator

public class DecisionTreeFunctionEstimator
extends org.processmining.datadiscovery.estimators.AbstractDecisionTreeFunctionEstimator

Field Summary

Fields

Modifier and Type	Field and Description
protected com.google.common.collect.BiMap<java.lang.Object,java.lang.Integer>	classIndexMap
protected com.google.common.collect.BiMap<java.lang.String,java.lang.Object>	classMapping
protected java.util.List<java.lang.String>	classValues
protected java.lang.Object[]	outputClasses

Fields inherited from class org.processmining.datadiscovery.estimators.AbstractDecisionTreeFunctionEstimator

attributeList, binarySplit, booleanValues, classAttributeName, confidenceThreshold, crossValidate, crossValidateRandom, evaluation, FALSE_VALUE, instances, minNumInstancePerLeaf, name, nullValue, numFoldCrossValidation, numFoldErrorPruning, saveData, tree, TRUE_VALUE, unpruned, variableType

Constructor Summary

Constructors

Constructor and Description
DecisionTreeFunctionEstimator(java.util.Map<java.lang.String,org.processmining.datadiscovery.estimators.Type> attributeType, java.util.Map<java.lang.String,java.util.Set<java.lang.String>> literalValues, java.lang.Object[] outputClasses, java.lang.String name, int capacity) Constructs a new DecisionTreeFunctionEstimator with the place's target transitions as CLASS value.

Method Summary

Modifier and Type	Method and Description
void	addInstance(java.util.Map<java.lang.String,java.lang.Object> variableAssignment, java.lang.Object outputValue, float weight) Adds a new instance to the estimator's 'instances'.
void	addWekaInstance(weka.core.Instance instance, java.lang.Object outputValue, float weight)
java.lang.Object	classifyInstance(java.util.Map<java.lang.String,java.lang.Object> attributes)
double	computeQualityMeasure() Evaluates the classification.
protected java.util.ArrayList<weka.core.Attribute>	createAttributeList(java.util.Map<java.lang.String,org.processmining.datadiscovery.estimators.Type> attributeType, java.util.Map<java.lang.String,java.util.Set<java.lang.String>> literalValues, java.lang.Object[] outputClasses) Create an ArrayList with all attributes used (including the class attribute)
protected weka.classifiers.AbstractClassifier	createClassifier(java.lang.Object[] option, boolean saveData) Creates the RepTree using the earlier supplied options.
weka.classifiers.AbstractClassifier	getClassifier()
com.google.common.collect.BiMap<java.lang.Object,java.lang.Integer>	getClassIndexMap()
protected java.lang.String	getClassValue(java.lang.Object outputValue) Retrieve the classValue from 'mapping' for a given target class (Object)
java.util.Map<java.lang.Object,org.processmining.datadiscovery.estimators.FunctionEstimation>	getFunctionEstimation(java.lang.Object[] option) Returns a mapping from a Transition
Instances	getInstances()
boolean	isTreatNoLeafAsFalse()
void	setTreatNoLeafAsFalse(boolean treatNoLeafAsFalse)

Methods inherited from class org.processmining.datadiscovery.estimators.AbstractDecisionTreeFunctionEstimator

classify, createAndSetTree, createInstance, getAttributeByName, getAttributeIndexMap, getConfidenceThreshold, getCrossValidateRandom, getEvaluation, getMinNumInstancePerLeaf, getName, getNumFoldCrossValidation, getNumFoldErrorPruning, getNumInstances, getPrefuseTreeVisualization, getSumOfWeights, getTreeClassificationAdapter, getVisualization, isBinarySplit, isCrossValidate, isUnpruned, saveInstances, setBinarySplit, setConfidenceFactor, setCrossValidate, setCrossValidateRandom, setMinNumObj, setNumFoldCrossValidation, setNumFolds, setSaveData, setUnpruned, toString

Methods inherited from class org.processmining.datadiscovery.estimators.AbstractFunctionEstimator

buildExpressionsFromLeafs

Methods inherited from class java.lang.Object

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

Field Detail

outputClasses

protected final java.lang.Object[] outputClasses

classValues

protected java.util.List<java.lang.String> classValues

classIndexMap

protected com.google.common.collect.BiMap<java.lang.Object,java.lang.Integer> classIndexMap

classMapping

protected com.google.common.collect.BiMap<java.lang.String,java.lang.Object> classMapping

Constructor Detail

DecisionTreeFunctionEstimator

public DecisionTreeFunctionEstimator(java.util.Map<java.lang.String,org.processmining.datadiscovery.estimators.Type> attributeType,
                                     java.util.Map<java.lang.String,java.util.Set<java.lang.String>> literalValues,
                                     java.lang.Object[] outputClasses,
                                     java.lang.String name,
                                     int capacity)

Constructs a new DecisionTreeFunctionEstimator with the place's target transitions as CLASS value.

Parameters:

attributeType - A Mapping from Attribute name String to Attribute Type (use AttributeUtil to create one from a XLog

literalValues - A Mapping from XAttributeLiteral attribute name String to a Set of literal values of type String (use AttributeUtil to create one from a XLog

outputClasses - An Object[] array of target CLASS objects.

name - The label of the place associated with this function estimator

capacity - parameter for weka 'instances'.

Method Detail

createAttributeList

protected java.util.ArrayList<weka.core.Attribute> createAttributeList(java.util.Map<java.lang.String,org.processmining.datadiscovery.estimators.Type> attributeType,
                                                                       java.util.Map<java.lang.String,java.util.Set<java.lang.String>> literalValues,
                                                                       java.lang.Object[] outputClasses)

Description copied from class: org.processmining.datadiscovery.estimators.AbstractDecisionTreeFunctionEstimator

Create an ArrayList with all attributes used (including the class attribute)

Specified by:

createAttributeList in class org.processmining.datadiscovery.estimators.AbstractDecisionTreeFunctionEstimator

Returns:

addInstance

public void addInstance(java.util.Map<java.lang.String,java.lang.Object> variableAssignment,
                        java.lang.Object outputValue,
                        float weight)

Adds a new instance to the estimator's 'instances'.

Parameters:

variableAssignment - A Map of variable identifier (key) and its value (object).

outputValue - The Transition which is to be executed with the variable values of variableAssignment.

weight - Parameter for weighted decision trees. Keep 1 for default.

addWekaInstance

public void addWekaInstance(weka.core.Instance instance,
                            java.lang.Object outputValue,
                            float weight)

getClassValue

protected java.lang.String getClassValue(java.lang.Object outputValue)

Retrieve the classValue from 'mapping' for a given target class (Object)

Parameters:

outputValue - A target class (Object)

Returns:

computeQualityMeasure

public double computeQualityMeasure()

Description copied from interface: org.processmining.datadiscovery.estimators.FunctionEstimator

Evaluates the classification.

Returns:

createClassifier

protected weka.classifiers.AbstractClassifier createClassifier(java.lang.Object[] option,
                                                               boolean saveData)
                                                        throws java.lang.Exception

Creates the RepTree using the earlier supplied options.

Specified by:

createClassifier in class org.processmining.datadiscovery.estimators.AbstractDecisionTreeFunctionEstimator

Parameters:

option - Array of Strings containing J48 tree options.

saveData - Boolean. True to enable saving instance data to the tree.

Returns:

a J48 tree defined by options, associated with 'instances'.

Throws:

java.lang.Exception - if classifier can't be built correctly

getFunctionEstimation

public java.util.Map<java.lang.Object,org.processmining.datadiscovery.estimators.FunctionEstimation> getFunctionEstimation(java.lang.Object[] option)
                                                                                                                    throws java.lang.Exception

Returns a mapping from a Transition