de.lmu.ifi.dbs.elki.evaluation.classification

Class ConfusionMatrix

java.lang.Object

de.lmu.ifi.dbs.elki.evaluation.classification.ConfusionMatrix

public class ConfusionMatrix
extends java.lang.Object

Provides a confusion matrix with some prediction performance measures that can be derived from a confusion matrix.

Since:

0.7.0

Author:

Arthur Zimek

Field Summary

Fields

Modifier and Type	Field and Description
private int[][]	confusion Holds the confusion matrix.
private java.util.ArrayList<ClassLabel>	labels Holds the class labels.

Constructor Summary

Constructors

Constructor and Description
ConfusionMatrix(java.util.ArrayList<ClassLabel> labels, int[][] confusion) Provides a confusion matrix for the given values.

Method Summary

Modifier and Type	Method and Description
int	colSum(int classindex) The number of instances present in the specified column.
int	falseNegatives(int classindex) The false negatives for the specified class.
double	falsePositiveRate() Provides the false positive rate.
double	falsePositiveRate(int classindex) Provides the false positive rate for the specified class.
int	falsePositives(int classindex) The false positives for the specified class.
double	positivePredictedValue() Provides the positive predicted value.
double	positivePredictedValue(int classindex) Provides the positive predicted value for the specified class.
int	rowSum(int classindex) The number of instances present in the specified row.
java.lang.String	toString() Provides a String representation of this confusion matrix.
int	totalInstances() The total number of instances covered by this confusion matrix.
int	trueNegatives(int classindex) The number of true negatives of the specified class.
double	truePositiveRate() Provides the true positive rate.
double	truePositiveRate(int classindex) Provides the true positive rate for the specified class.
int	truePositives() The number of correctly classified instances.
int	truePositives(int classindex) The number of correctly classified instances belonging to the specified class.
int	value(int trueClassindex, int predictedClassindex) The number of instances belonging to class trueClassindex and predicted as predictedClassindex.

Methods inherited from class java.lang.Object

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

Field Detail

confusion

private int[][] confusion

Holds the confusion matrix. Must be a square matrix. The rows (first index) give the values which classes are classified as row, the columns (second index) give the values the class col has been classified as. Thus, confusion[predicted][real] addresses the number of instances of class real that have been assigned to the class predicted.

labels

private java.util.ArrayList<ClassLabel> labels

Holds the class labels.

Constructor Detail

ConfusionMatrix

public ConfusionMatrix(java.util.ArrayList<ClassLabel> labels,
                       int[][] confusion)
                throws java.lang.IllegalArgumentException

Provides a confusion matrix for the given values.

Parameters:

labels - the class labels - must conform the confusion matrix in length

confusion - the confusion matrix. Must be a square matrix. The rows (first index) give the values which classes are classified as row, the columns (second index) give the values the class col has been classified as. Thus, confusion[predicted][real] addresses the number of instances of class real that have been assigned to the class predicted.

Throws:

java.lang.IllegalArgumentException - if the confusion matrix is not square or not complete or if the length of class labels does not conform the length of the confusion matrix

Method Detail

truePositiveRate

public double truePositiveRate()

Provides the true positive rate. Aka accuracy or sensitivity or recall: TP / (TP+FN).

Returns:

the true positive rate

falsePositiveRate

public double falsePositiveRate(int classindex)

Provides the false positive rate for the specified class.

Parameters:

classindex - the index of the class to retrieve the false positive rate for

Returns:

the false positive rate for the specified class

falsePositiveRate

public double falsePositiveRate()

Provides the false positive rate. Aka false alarm rate: FP / (FP+TN).

Returns:

the false positive rate

positivePredictedValue

public double positivePredictedValue(int classindex)

Provides the positive predicted value for the specified class.

Parameters:

classindex - the index of the class to retrieve the positive predicted value for

Returns:

the positive predicted value for the specified class

positivePredictedValue

public double positivePredictedValue()

Provides the positive predicted value. Aka precision or specificity: TP / (TP+FP).

Returns:

the positive predicted value

truePositives

public int truePositives()

The number of correctly classified instances.

Returns:

the number of correctly classified instances

truePositives

public int truePositives(int classindex)

The number of correctly classified instances belonging to the specified class.

Parameters:

classindex - the index of the class to retrieve the correctly classified instances of

Returns:

the number of correctly classified instances belonging to the specified class

truePositiveRate

public double truePositiveRate(int classindex)

Provides the true positive rate for the specified class.

Parameters:

classindex - the index of the class to retrieve the true positive rate for

Returns:

the true positive rate

trueNegatives

public int trueNegatives(int classindex)

The number of true negatives of the specified class.

Parameters:

classindex - the index of the class to retrieve the true negatives for

Returns:

the number of true negatives of the specified class

falsePositives

public int falsePositives(int classindex)

The false positives for the specified class.

Parameters:

classindex - the index of the class to retrieve the false positives for

Returns:

the false positives for the specified class

falseNegatives

public int falseNegatives(int classindex)

The false negatives for the specified class.

Parameters:

classindex - the index of the class to retrieve the false negatives for

Returns:

the false negatives for the specified class

totalInstances

public int totalInstances()

The total number of instances covered by this confusion matrix.

Returns:

the total number of instances covered by this confusion matrix

rowSum

public int rowSum(int classindex)

The number of instances present in the specified row. I.e., classified as class classindex.

Parameters:

classindex - the index of the class the resulting number of instances has been classified as

Returns:

the number of instances present in the specified row

colSum

public int colSum(int classindex)

The number of instances present in the specified column. I.e., the instances of class classindex.

Parameters:

classindex - the index of the class theresulting number of instances belongs to

Returns:

the number of instances present in the specified column

value

public int value(int trueClassindex,
                 int predictedClassindex)

The number of instances belonging to class trueClassindex and predicted as predictedClassindex.

Parameters:

trueClassindex - the true class index

predictedClassindex - the predicted class index

Returns:

the number of instances belonging to class trueClassindex and predicted as predictedClassindex

toString

public java.lang.String toString()

Provides a String representation of this confusion matrix.

Overrides:

toString in class java.lang.Object

See Also:

Object.toString()