org.tribuo.classification.evaluation

Class ConfusionMetrics

java.lang.Object

org.tribuo.classification.evaluation.ConfusionMetrics

public final class ConfusionMetrics
extends Object

Static functions for computing classification metrics based on a ConfusionMatrix.

Method Summary

Modifier and Type	Method and Description
static <T extends Classifiable<T>> double	accuracy(EvaluationMetric.Average average, ConfusionMatrix<T> cm) Calculates the accuracy using the specified average type and confusion matrix.
static <T extends Classifiable<T>> double	accuracy(MetricTarget<T> target, ConfusionMatrix<T> cm) Calculates the accuracy given this confusion matrix.
static <T extends Classifiable<T>> double	accuracy(T label, ConfusionMatrix<T> cm) Calculates a per label accuracy given this confusion matrix.
static <T extends Classifiable<T>> double	balancedErrorRate(ConfusionMatrix<T> cm) Calculates the balanced error rate, i.e., the mean of the recalls.
static double	f1(double tp, double fp, double tn, double fn) Computes the F_1 score.
static <T extends Classifiable<T>> double	f1(MetricTarget<T> tgt, ConfusionMatrix<T> cm) Computes the F_1 score.
static <T extends Classifiable<T>> double	fn(MetricTarget<T> tgt, ConfusionMatrix<T> cm) Returns the number of false negatives, possibly averaged depending on the metric target.
static <T extends Classifiable<T>> double	fp(MetricTarget<T> tgt, ConfusionMatrix<T> cm) Returns the number of false positives, possibly averaged depending on the metric target.
static double	fscore(double beta, double tp, double fp, double tn, double fn) Computes the Fscore.
static <T extends Classifiable<T>> double	fscore(MetricTarget<T> tgt, ConfusionMatrix<T> cm, double beta) Computes the Fscore.
static double	precision(double tp, double fp, double tn, double fn) Calculates the precision based upon the supplied statistics.
static <T extends Classifiable<T>> double	precision(MetricTarget<T> tgt, ConfusionMatrix<T> cm) Calculates the precision for this metric target.
static double	recall(double tp, double fp, double tn, double fn) Calculates the recall based upon the supplied statistics.
static <T extends Classifiable<T>> double	recall(MetricTarget<T> tgt, ConfusionMatrix<T> cm) Calculates the recall for this metric target.
static <T extends Classifiable<T>> double	tn(MetricTarget<T> tgt, ConfusionMatrix<T> cm) Returns the number of true negatives, possibly averaged depending on the metric target.
static <T extends Classifiable<T>> double	tp(MetricTarget<T> tgt, ConfusionMatrix<T> cm) Returns the number of true positives, possibly averaged depending on the metric target.

Methods inherited from class java.lang.Object

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

Method Detail

accuracy

public static <T extends Classifiable<T>> double accuracy(MetricTarget<T> target,
                                                          ConfusionMatrix<T> cm)

Calculates the accuracy given this confusion matrix.

Type Parameters:

T - The type parameter

Parameters:

target - The metric target

cm - The confusion matrix

Returns:

The accuracy

accuracy

public static <T extends Classifiable<T>> double accuracy(T label,
                                                          ConfusionMatrix<T> cm)

Calculates a per label accuracy given this confusion matrix.

Type Parameters:

T - The type parameter

Parameters:

label - The label

cm - The confusion matrix

Returns:

The accuracy

accuracy

public static <T extends Classifiable<T>> double accuracy(EvaluationMetric.Average average,
                                                          ConfusionMatrix<T> cm)

Calculates the accuracy using the specified average type and confusion matrix.

Type Parameters:

T - the type parameter

Parameters:

average - the average

cm - The confusion matrix

Returns:

The accuracy

balancedErrorRate

public static <T extends Classifiable<T>> double balancedErrorRate(ConfusionMatrix<T> cm)

Calculates the balanced error rate, i.e., the mean of the recalls.

Type Parameters:

T - the type parameter

Parameters:

cm - The confusion matrix

Returns:

the balanced error rate.

tp

public static <T extends Classifiable<T>> double tp(MetricTarget<T> tgt,
                                                    ConfusionMatrix<T> cm)

Returns the number of true positives, possibly averaged depending on the metric target.

Type Parameters:

T - the type parameter

Parameters:

tgt - The metric target

cm - The confusion matrix

Returns:

the true positives.

fp

public static <T extends Classifiable<T>> double fp(MetricTarget<T> tgt,
                                                    ConfusionMatrix<T> cm)

Returns the number of false positives, possibly averaged depending on the metric target.

Type Parameters:

T - the type parameter

Parameters:

tgt - The metric target

cm - The confusion matrix

Returns:

the false positives.

tn

public static <T extends Classifiable<T>> double tn(MetricTarget<T> tgt,
                                                    ConfusionMatrix<T> cm)

Returns the number of true negatives, possibly averaged depending on the metric target.

Type Parameters:

T - the type parameter

Parameters:

tgt - The metric target

cm - The confusion matrix

Returns:

the true negatives.

fn

public static <T extends Classifiable<T>> double fn(MetricTarget<T> tgt,
                                                    ConfusionMatrix<T> cm)

Returns the number of false negatives, possibly averaged depending on the metric target.

Type Parameters:

T - the type parameter

Parameters:

tgt - The metric target

cm - The confusion matrix

Returns:

the false negatives.

precision

public static <T extends Classifiable<T>> double precision(MetricTarget<T> tgt,
                                                           ConfusionMatrix<T> cm)

Calculates the precision for this metric target.

Type Parameters:

T - the type parameter

Parameters:

tgt - The metric target

cm - The confusion matrix

Returns:

the precision.

precision

public static double precision(double tp,
                               double fp,
                               double tn,
                               double fn)

Calculates the precision based upon the supplied statistics.

Parameters:

tp - the true positives

fp - the false positives

tn - the true negatives

fn - the false negatives

Returns:

The recall.

recall

public static <T extends Classifiable<T>> double recall(MetricTarget<T> tgt,
                                                        ConfusionMatrix<T> cm)

Calculates the recall for this metric target.

Type Parameters:

T - the type parameter

Parameters:

tgt - The metric target

cm - The confusion matrix

Returns:

The recall.

recall

public static double recall(double tp,
                            double fp,
                            double tn,
                            double fn)

Calculates the recall based upon the supplied statistics.

Parameters:

tp - the true positives

fp - the false positives

tn - the true negatives

fn - the false negatives

Returns:

The recall.

f1

public static <T extends Classifiable<T>> double f1(MetricTarget<T> tgt,
                                                    ConfusionMatrix<T> cm)

Computes the F_1 score.

Type Parameters:

T - the type parameter

Parameters:

tgt - the metric target.

cm - the confusion matrix.

Returns:

the F_1 score.

f1

public static double f1(double tp,
                        double fp,
                        double tn,
                        double fn)

Computes the F_1 score.

Parameters:

tp - the true positives

fp - the false positives

tn - the true negatives

fn - the false negatives

Returns:

the F_1 score.

fscore

public static double fscore(double beta,
                            double tp,
                            double fp,
                            double tn,
                            double fn)

Computes the Fscore.

Parameters:

beta - the beta.

tp - the true positives.

fp - the false positives.

tn - the true negatives.

fn - the false negatives.

Returns:

the F_beta score.

fscore

public static <T extends Classifiable<T>> double fscore(MetricTarget<T> tgt,
                                                        ConfusionMatrix<T> cm,
                                                        double beta)

Computes the Fscore.

Type Parameters:

T - the type parameter

Parameters:

tgt - The metric target

cm - The confusion matrix

beta - the beta

Returns:

The F_beta score.