org.deeplearning4j.eval

Class ROCMultiClass

java.lang.Object

org.deeplearning4j.eval.BaseEvaluation<ROCMultiClass>

org.deeplearning4j.eval.ROCMultiClass

All Implemented Interfaces:

Serializable, IEvaluation<ROCMultiClass>

public class ROCMultiClass
extends BaseEvaluation<ROCMultiClass>

ROC (Receiver Operating Characteristic) for multi-class classifiers. As per ROC, ROCBinary supports both exact (thersholdSteps == 0) and thresholded; see ROC for details.

The ROC curves are produced by treating the predictions as a set of one-vs-all classifiers, and then calculating ROC curves for each. In practice, this means for N classes, we get N ROC curves.

Author:

Alex Black

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
static int	DEFAULT_STATS_PRECISION

Constructor Summary

Constructors

Constructor and Description
ROCMultiClass()
ROCMultiClass(int thresholdSteps)
ROCMultiClass(int thresholdSteps, boolean rocRemoveRedundantPts)

Method Summary

Modifier and Type	Method and Description
double	calculateAUC(int classIdx) Calculate the AUC - Area Under ROC Curve Utilizes trapezoidal integration internally
double	calculateAUCPR(int classIdx) Calculate the AUPRC - Area Under Curve Precision Recall Utilizes trapezoidal integration internally
double	calculateAverageAUC() Calculate the macro-average (one-vs-all) AUC for all classes
void	eval(org.nd4j.linalg.api.ndarray.INDArray labels, org.nd4j.linalg.api.ndarray.INDArray predictions) Evaluate (collect statistics for) the given minibatch of data.
long	getCountActualNegative(int outputNum) Get the actual negative count (accounting for any masking) for the specified output/column
long	getCountActualPositive(int outputNum) Get the actual positive count (accounting for any masking) for the specified class
int	getNumClasses()
PrecisionRecallCurve	getPrecisionRecallCurve(int classIdx) Get the (one vs.
RocCurve	getRocCurve(int classIdx) Get the (one vs.
void	merge(ROCMultiClass other) Merge this ROCMultiClass instance with another.
void	reset()
String	stats()
String	stats(int printPrecision)

Methods inherited from class org.deeplearning4j.eval.BaseEvaluation

equals, eval, eval, evalTimeSeries, evalTimeSeries, fromJson, fromYaml, toJson, toString, toYaml

Methods inherited from class java.lang.Object

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

Field Detail

DEFAULT_STATS_PRECISION

public static final int DEFAULT_STATS_PRECISION

See Also:

Constant Field Values

Constructor Detail

ROCMultiClass

public ROCMultiClass()

ROCMultiClass

public ROCMultiClass(int thresholdSteps)

Parameters:

thresholdSteps - Number of threshold steps to use for the ROC calculation. Set to 0 for exact ROC calculation

ROCMultiClass

public ROCMultiClass(int thresholdSteps,
                     boolean rocRemoveRedundantPts)

Parameters:

thresholdSteps - Number of threshold steps to use for the ROC calculation. If set to 0: use exact calculation

rocRemoveRedundantPts - Usually set to true. If true, remove any redundant points from ROC and P-R curves

Method Detail

reset

public void reset()

stats

public String stats()

Returns:

stats

public String stats(int printPrecision)

eval

public void eval(org.nd4j.linalg.api.ndarray.INDArray labels,
                 org.nd4j.linalg.api.ndarray.INDArray predictions)

Evaluate (collect statistics for) the given minibatch of data. For time series (3 dimensions) use BaseEvaluation.evalTimeSeries(INDArray, INDArray) or BaseEvaluation.evalTimeSeries(INDArray, INDArray, INDArray)

Parameters:

labels - Labels / true outcomes

predictions - Predictions

getRocCurve

public RocCurve getRocCurve(int classIdx)

Get the (one vs. all) ROC curve for the specified class

Parameters:

classIdx - Class index to get the ROC curve for

Returns:

ROC curve for the given class

getPrecisionRecallCurve

public PrecisionRecallCurve getPrecisionRecallCurve(int classIdx)

Get the (one vs. all) Precision-Recall curve for the specified class

Parameters:

classIdx - Class to get the P-R curve for

Returns:

Precision recall curve for the given class

calculateAUC

public double calculateAUC(int classIdx)

Calculate the AUC - Area Under ROC Curve
Utilizes trapezoidal integration internally

Returns:

AUC

calculateAUCPR

public double calculateAUCPR(int classIdx)

Calculate the AUPRC - Area Under Curve Precision Recall
Utilizes trapezoidal integration internally

Returns:

AUC

calculateAverageAUC

public double calculateAverageAUC()

Calculate the macro-average (one-vs-all) AUC for all classes

getCountActualPositive

public long getCountActualPositive(int outputNum)

Get the actual positive count (accounting for any masking) for the specified class

Parameters:

outputNum - Index of the class

getCountActualNegative

public long getCountActualNegative(int outputNum)

Get the actual negative count (accounting for any masking) for the specified output/column

Parameters:

outputNum - Index of the class

merge

public void merge(ROCMultiClass other)

Merge this ROCMultiClass instance with another. This ROCMultiClass instance is modified, by adding the stats from the other instance.

Parameters:

other - ROCMultiClass instance to combine with this one

getNumClasses

public int getNumClasses()