gov.sandia.cognition.statistics.method

Class ReceiverOperatingCharacteristic.Statistic

java.lang.Object

gov.sandia.cognition.util.AbstractCloneableSerializable

gov.sandia.cognition.statistics.method.AbstractConfidenceStatistic

gov.sandia.cognition.statistics.method.MannWhitneyUConfidence.Statistic

gov.sandia.cognition.statistics.method.ReceiverOperatingCharacteristic.Statistic

All Implemented Interfaces:

ConfidenceStatistic, CloneableSerializable, java.io.Serializable, java.lang.Cloneable

Enclosing class:

ReceiverOperatingCharacteristic

public static class ReceiverOperatingCharacteristic.Statistic
extends MannWhitneyUConfidence.Statistic

Contains useful statistics derived from a ROC curve

See Also:

Serialized Form

Field Summary

Fields inherited from class gov.sandia.cognition.statistics.method.AbstractConfidenceStatistic

nullHypothesisProbability

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	Statistic(ReceiverOperatingCharacteristic roc) Creates a new instance of Statistic

Method Summary

Modifier and Type	Method and Description
static double	computeAreaUnderCurve(ReceiverOperatingCharacteristic roc) Computes the "pessimistic" area under the ROC curve using the top-left rectangle method for numerical integration.
static double	computeAreaUnderCurveTopLeft(java.util.Collection<ReceiverOperatingCharacteristic.DataPoint> points) Computes the Area Under Curve for an x-axis sorted Collection of ROC points using the top-left rectangle method for numerical integration.
static double	computeAreaUnderCurveTrapezoid(java.util.Collection<ReceiverOperatingCharacteristic.DataPoint> points) Computes the Area Under Curve for an x-axis sorted Collection of ROC points using the top-left rectangle method for numerical integration.
static double	computeDPrime(ReceiverOperatingCharacteristic.DataPoint data) Computes the value of d-prime given a datapoint
static ReceiverOperatingCharacteristic.DataPoint	computeOptimalThreshold(ReceiverOperatingCharacteristic roc) Determines the DataPoint, and associated threshold, that simultaneously maximizes the value of Area=TruePositiveRate+TrueNegativeRate, usually the upper-left "knee" on the ROC curve.
static ReceiverOperatingCharacteristic.DataPoint	computeOptimalThreshold(ReceiverOperatingCharacteristic roc, double truePositiveWeight, double trueNegativeWeight) Determines the DataPoint, and associated threshold, that simultaneously maximizes the value of Area=TruePositiveRate+TrueNegativeRate, usually the upper-left "knee" on the ROC curve.
double	getAreaUnderCurve() Getter for areaUnderCurve
double	getDPrime() Getter for dPrime
ReceiverOperatingCharacteristic.DataPoint	getOptimalThreshold() Getter for optimalThreshold
protected void	setAreaUnderCurve(double areaUnderCurve) Setter for areaUnderCurve
protected void	setDPrime(double dPrime) Setter for dPrime
protected void	setOptimalThreshold(ReceiverOperatingCharacteristic.DataPoint optimalThreshold) Setter for optimalThreshold

Methods inherited from class gov.sandia.cognition.statistics.method.MannWhitneyUConfidence.Statistic

computeNullHypothesisProbability, computeU, computeZ, getN1, getN2, getTestStatistic, getU, getZ, setN1, setN2, setU, setZ

Methods inherited from class gov.sandia.cognition.statistics.method.AbstractConfidenceStatistic

getNullHypothesisProbability, setNullHypothesisProbability, toString

Methods inherited from class gov.sandia.cognition.util.AbstractCloneableSerializable

clone

Methods inherited from class java.lang.Object

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

Methods inherited from interface gov.sandia.cognition.util.CloneableSerializable

clone

Constructor Detail

Statistic

protected Statistic(ReceiverOperatingCharacteristic roc)

Creates a new instance of Statistic

Parameters:

roc - ROC Curve from which to pull statistics

Method Detail

computeAreaUnderCurve

public static double computeAreaUnderCurve(ReceiverOperatingCharacteristic roc)

Computes the "pessimistic" area under the ROC curve using the top-left rectangle method for numerical integration.

Parameters:

roc - ROC Curve to compute the area under

Returns:

Area underneath the ROC curve, on the interval [0,1]. A value of 0.5 means that the classifier is doing no better than chance and bigger is better

computeAreaUnderCurveTopLeft

@PublicationReference(author="Wikipedia",
                      title="Rectangle method",
                      type=WebPage,
                      year=2011,
                      url="http://en.wikipedia.org/wiki/Rectangle_method")
public static double computeAreaUnderCurveTopLeft(java.util.Collection<ReceiverOperatingCharacteristic.DataPoint> points)

Computes the Area Under Curve for an x-axis sorted Collection of ROC points using the top-left rectangle method for numerical integration.

Parameters:

points - x-axis sorted collection of x-axis points

Returns:

Area underneath the ROC curve, on the interval [0,1]. A value of 0.5 means that the classifier is doing no better than chance and bigger is better

computeAreaUnderCurveTrapezoid

@PublicationReference(author="Wikipedia",
                      title="Trapezoidal rule",
                      type=WebPage,
                      year=2011,
                      url="http://en.wikipedia.org/wiki/Trapezoidal_rule")
public static double computeAreaUnderCurveTrapezoid(java.util.Collection<ReceiverOperatingCharacteristic.DataPoint> points)

Computes the Area Under Curve for an x-axis sorted Collection of ROC points using the top-left rectangle method for numerical integration.

Parameters:

points - x-axis sorted collection of x-axis points

Returns:

Area underneath the ROC curve, on the interval [0,1]. A value of 0.5 means that the classifier is doing no better than chance and bigger is better

computeOptimalThreshold

public static ReceiverOperatingCharacteristic.DataPoint computeOptimalThreshold(ReceiverOperatingCharacteristic roc)

Determines the DataPoint, and associated threshold, that simultaneously maximizes the value of Area=TruePositiveRate+TrueNegativeRate, usually the upper-left "knee" on the ROC curve.

Parameters:

roc - ROC Curve to consider

Returns:

DataPoint, with corresponding threshold, that maximizes the value of Area=TruePositiveRate*(1-FalsePositiveRate), usually the upper-left "knee" on the ROC curve.

computeOptimalThreshold

public static ReceiverOperatingCharacteristic.DataPoint computeOptimalThreshold(ReceiverOperatingCharacteristic roc,
                                                                                double truePositiveWeight,
                                                                                double trueNegativeWeight)

Determines the DataPoint, and associated threshold, that simultaneously maximizes the value of Area=TruePositiveRate+TrueNegativeRate, usually the upper-left "knee" on the ROC curve.

Parameters:

truePositiveWeight - Amount to weight the TruePositiveRate

trueNegativeWeight - Amount to weight the TrueNegativeRate

roc - ROC Curve to consider

Returns:

DataPoint, with corresponding threshold, that maximizes the value of Area=TruePositiveRate*(1-FalsePositiveRate), usually the upper-left "knee" on the ROC curve.

computeDPrime

public static double computeDPrime(ReceiverOperatingCharacteristic.DataPoint data)

Computes the value of d-prime given a datapoint

Parameters:

data - Datapoint from which to estimate d'

Returns:

Estimated distance between the two classes to be split. Larger values of d' indicate that the classes are easier to split, d'=0 means that the classes overlap, and negative values mean that your classifier is doing worse than chance, chump. This appears to only be used by psychologists.

getDPrime

public double getDPrime()

Getter for dPrime

Returns:

Estimated distance between the two classes to be split. Larger values of d' indicate that the classes are easier to split, d'=0 means that the classes overlap, and negative values mean that your classifier is doing worse than chance, chump. This appears to only be used by psychologists.

setDPrime

protected void setDPrime(double dPrime)

Setter for dPrime

Parameters:

dPrime - Estimated distance between the two classes to be split. Larger values of d' indicate that the classes are easier to split, d'=0 means that the classes overlap, and negative values mean that your classifier is doing worse than chance, chump. This appears to only be used by psychologists.

getAreaUnderCurve

public double getAreaUnderCurve()

Getter for areaUnderCurve

Returns:

Area underneath the ROC curve, on the interval [0,1]. A value of 0.5 means that the classifier is doing no better than chance and bigger is better

setAreaUnderCurve

protected void setAreaUnderCurve(double areaUnderCurve)

Setter for areaUnderCurve

Parameters:

areaUnderCurve - Area underneath the ROC curve, on the interval [0,1]. A value of 0.5 means that the classifier is doing no better than chance and bigger is better

getOptimalThreshold

public ReceiverOperatingCharacteristic.DataPoint getOptimalThreshold()

Getter for optimalThreshold

Returns:

DataPoint, with corresponding threshold, that maximizes the value of Area=TruePositiveRate*(1-FalsePositiveRate), usually the upper-left "knee" on the ROC curve.

setOptimalThreshold

protected void setOptimalThreshold(ReceiverOperatingCharacteristic.DataPoint optimalThreshold)

Setter for optimalThreshold

Parameters:

optimalThreshold - DataPoint, with corresponding threshold, that maximizes the value of Area=TruePositiveRate*(1-FalsePositiveRate), usually the upper-left "knee" on the ROC curve.