de.lmu.ifi.dbs.elki.evaluation
Class ComputeROCCurve<O extends DatabaseObject>

java.lang.Object
  de.lmu.ifi.dbs.elki.logging.AbstractLoggable
      de.lmu.ifi.dbs.elki.utilities.optionhandling.AbstractParameterizable
          de.lmu.ifi.dbs.elki.algorithm.AbstractAlgorithm<O,MultiResult>
              de.lmu.ifi.dbs.elki.evaluation.ComputeROCCurve<O>

Type Parameters:

O - Database object type

All Implemented Interfaces:

Algorithm<O,MultiResult>, Parameterizable

public class ComputeROCCurve<O extends DatabaseObject>
extends AbstractAlgorithm<O,MultiResult>

Compute a ROC curve to evaluate a ranking algorithm and compute the corresponding ROCAUC value. The parameter -rocauc.positive specifies the class label of "positive" hits. The nested algorithm -algorithm will be run, the result will be searched for an iterable or ordering result, which then is compared with the clustering obtained via the given class label.

Author:

Erich Schubert

Field Summary

private Algorithm<O,Result>	algorithm Holds the algorithm to run.
private ClassParameter<Algorithm<O,Result>>	ALGORITHM_PARAM Parameter to specify the algorithm to be applied, must extend Algorithm.
private String	positive_class_name Stores the "positive" class.
static OptionID	POSITIVE_CLASS_NAME_ID OptionID for POSITIVE_CLASS_NAME_PARAM
private PatternParameter	POSITIVE_CLASS_NAME_PARAM The distance function to determine the reachability distance between database objects.
private MultiResult	result Stores the result object.
static AssociationID<Double>	ROC_AUC The association id to associate the ROC Area-under-Curve.

Fields inherited from class de.lmu.ifi.dbs.elki.utilities.optionhandling.AbstractParameterizable

optionHandler

Fields inherited from class de.lmu.ifi.dbs.elki.logging.AbstractLoggable

debug, logger

Constructor Summary

ComputeROCCurve()

Method Summary

Description	getDescription() Returns a description of the algorithm.
private Iterator<Integer>	getIterableResult(Database<O> database, Result result) Find an "iterable" result that looks like object IDs.
private Cluster<Model>	getReferenceCluster(Database<O> database, String class_name) Find the "positive" reference cluster using a by label clustering.
MultiResult	getResult() Returns the result of the algorithm.
protected MultiResult	runInTime(Database<O> database) The run method encapsulated in measure of runtime.
List<String>	setParameters(List<String> args) Calls the super method and sets additionally the values of the flags AbstractAlgorithm.VERBOSE_FLAG and AbstractAlgorithm.TIME_FLAG.

Methods inherited from class de.lmu.ifi.dbs.elki.algorithm.AbstractAlgorithm

isTime, isVerbose, run, setTime, setVerbose

Methods inherited from class de.lmu.ifi.dbs.elki.utilities.optionhandling.AbstractParameterizable

addOption, addParameterizable, addParameterizable, checkGlobalParameterConstraints, collectOptions, getAttributeSettings, getParameters, rememberParametersExcept, removeOption, removeParameterizable, shortDescription

Methods inherited from class de.lmu.ifi.dbs.elki.logging.AbstractLoggable

debugFine, debugFiner, debugFinest, exception, progress, verbose, warning

Methods inherited from class java.lang.Object

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

Methods inherited from interface de.lmu.ifi.dbs.elki.utilities.optionhandling.Parameterizable

checkGlobalParameterConstraints, collectOptions, getParameters, shortDescription

Field Detail

POSITIVE_CLASS_NAME_ID

public static final OptionID POSITIVE_CLASS_NAME_ID

OptionID for POSITIVE_CLASS_NAME_PARAM

POSITIVE_CLASS_NAME_PARAM

private final PatternParameter POSITIVE_CLASS_NAME_PARAM

The distance function to determine the reachability distance between database objects.

Default value: EuclideanDistanceFunction

Key: -rocauc.positive

ALGORITHM_PARAM

private final ClassParameter<Algorithm<O extends DatabaseObject,Result>> ALGORITHM_PARAM

Parameter to specify the algorithm to be applied, must extend Algorithm.

Key: -algorithm

positive_class_name

private String positive_class_name

Stores the "positive" class.

algorithm

private Algorithm<O extends DatabaseObject,Result> algorithm

Holds the algorithm to run.

ROC_AUC

public static final AssociationID<Double> ROC_AUC

The association id to associate the ROC Area-under-Curve.

result

private MultiResult result

Stores the result object.

Constructor Detail

ComputeROCCurve

public ComputeROCCurve()

Method Detail

runInTime

protected MultiResult runInTime(Database<O> database)
                         throws IllegalStateException

Description copied from class: AbstractAlgorithm

The run method encapsulated in measure of runtime. An extending class needs not to take care of runtime itself.

Specified by:

runInTime in class AbstractAlgorithm<O extends DatabaseObject,MultiResult>

Parameters:

database - the database to run the algorithm on

Returns:

the Result computed by this algorithm

Throws:

IllegalStateException - if the algorithm has not been initialized properly (e.g. the setParameters(String[]) method has been failed to be called).

getReferenceCluster

private Cluster<Model> getReferenceCluster(Database<O> database,
                                           String class_name)

Find the "positive" reference cluster using a by label clustering.

Parameters:

database - Database to search in

class_name - Cluster name

Returns:

found cluster or it throws an exception.

getIterableResult

private Iterator<Integer> getIterableResult(Database<O> database,
                                            Result result)

Find an "iterable" result that looks like object IDs.

Parameters:

database - Database context

result - Result object

Returns:

Iterator to work with

getDescription

public Description getDescription()

Description copied from interface: Algorithm

Returns a description of the algorithm.

Returns:

a description of the algorithm

getResult

public MultiResult getResult()

Description copied from interface: Algorithm

Returns the result of the algorithm.

Returns:

the result of the algorithm

setParameters

public List<String> setParameters(List<String> args)
                           throws ParameterException

Description copied from class: AbstractAlgorithm

Calls the super method and sets additionally the values of the flags AbstractAlgorithm.VERBOSE_FLAG and AbstractAlgorithm.TIME_FLAG.

Specified by:

setParameters in interface Parameterizable

Overrides:

setParameters in class AbstractAlgorithm<O extends DatabaseObject,MultiResult>

Parameters:

args - parameters to set the attributes accordingly to

Returns:

a list containing the unused parameters

Throws:

ParameterException - in case of wrong parameter-setting