Options
All
  • Public
  • Public/Protected
  • All
Menu

Class ConfusionMatrix<T>

Constructs a confusion matrix

example

const CM = new ConfusionMatrix([[13, 2], [10, 5]], ['cat', 'dog'])

param matrix

The confusion matrix, a 2D Array. Rows represent the actual label and columns the predicted label.

param labels

Labels of the confusion matrix, a 1D Array

Type parameters

  • T: Label

Hierarchy

  • ConfusionMatrix

Index

Constructors

Accessors

Methods

Constructors

constructor

  • new ConfusionMatrix<T>(matrix: number[][], labels: T[]): ConfusionMatrix<T>

Accessors

accuracy

  • get accuracy(): number

  • Compute the general prediction accuracy

    deprecated

    Use getAccuracy

    Returns number

    The prediction accuracy ([0-1]

total

  • get total(): number

  • Compute the number of predicted observations

    deprecated

    Use getTotalCount

    Returns number

Methods

getAccuracy

  • getAccuracy(): number

  • Get total accuracy. The ratio between the number of true predictions and total number of classifications ([0-1])

    Returns number

getConfusionTable

  • getConfusionTable(label: T): number[][]

  • Get the confusion table.

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number[][]

    The 2x2 confusion table. [[TP, FN], [FP, TN]]

getCount

  • getCount(actual: T, predicted: T): number

  • Returns the element in the confusion matrix that corresponds to the given actual and predicted labels.

    Parameters

    • actual: T

      The true label

    • predicted: T

      The predicted label

    Returns number

    The element in the confusion matrix

getF1Score

  • getF1Score(label: T): number

getFalseCount

  • getFalseCount(): number

  • Get the total number of false predictions.

    Returns number

getFalseDiscoveryRate

  • getFalseDiscoveryRate(label: T): number

getFalseNegativeCount

  • getFalseNegativeCount(label: T): number

  • Get the number of false negative predictions.

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number

getFalseNegativeRate

  • getFalseNegativeRate(label: T): number

getFalseOmissionRate

  • getFalseOmissionRate(label: T): number

  • False omission rate (FOR)

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number

getFalsePositiveCount

  • getFalsePositiveCount(label: T): number

  • Get the number of false positive predictions.

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number

getFalsePositiveRate

  • getFalsePositiveRate(label: T): number

getIndex

  • getIndex(label: T): number

  • Get the index in the confusion matrix that corresponds to the given label

    throws

    if the label is not found

    Parameters

    • label: T

      The label to search for

    Returns number

getInformedness

  • getInformedness(label: T): number

getLabels

  • getLabels(): T[]

getMarkedness

  • getMarkedness(label: T): number

  • Markedness

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number

getMatrix

  • getMatrix(): number[][]

  • Get the confusion matrix

    Returns number[][]

getMatthewsCorrelationCoefficient

  • getMatthewsCorrelationCoefficient(label: T): number

getNegativeCount

  • getNegativeCount(label: T): number

  • Get the number of real negative samples.

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number

getNegativePredictiveValue

  • getNegativePredictiveValue(label: T): number

getPositiveCount

  • getPositiveCount(label: T): number

  • Get the number of real positive samples.

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number

getPositivePredictiveValue

  • getPositivePredictiveValue(label: T): number

getTotalCount

  • getTotalCount(): number

  • Get the total number of samples

    Returns number

getTrueCount

  • getTrueCount(): number

  • Get the total number of true predictions

    Returns number

getTrueNegativeCount

  • getTrueNegativeCount(label: T): number

  • Get the number of true negative predictions.

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number

getTrueNegativeRate

  • getTrueNegativeRate(label: T): number

  • Get the true negative rate a.k.a. specificity. Computes the ration between the number of true negative predictions and the total number of negative samples. https://en.wikipedia.org/wiki/Sensitivity_and_specificity

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number

    The true negative rate a.k.a. specificity.

getTruePositiveCount

  • getTruePositiveCount(label: T): number

  • Get the number of true positive predictions.

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number

getTruePositiveRate

  • getTruePositiveRate(label: T): number

  • Get the true positive rate a.k.a. sensitivity. Computes the ratio between the number of true positive predictions and the total number of positive samples. https://en.wikipedia.org/wiki/Sensitivity_and_specificity

    Parameters

    • label: T

      The label that should be considered "positive"

    Returns number

    The true positive rate [0-1]

Static fromLabels

  • fromLabels<T>(actual: T[], predicted: T[], options?: FromLabelsOptions<T>): ConfusionMatrix<T>

  • Construct confusion matrix from the predicted and actual labels (classes). Be sure to provide the arguments in the correct order!

    Type parameters

    • T: Label

    Parameters

    • actual: T[]

      The predicted labels of the classification

    • predicted: T[]

      The actual labels of the classification. Has to be of same length as predicted.

    • options: FromLabelsOptions<T> = {}

    Returns ConfusionMatrix<T>

    Confusion matrix