ConfusionMatrix (Confusion matrix 1.0 API)

java.lang.Object
- com.github.habernal.confusionmatrix.ConfusionMatrix

```
public class ConfusionMatrix
extends Object
```
Implementation of confusion matrix for evaluating learning algorithms; computes macro F-measure, accuracy, confidence intervals, Cohen's Kappa

Author:

Ivan Habernal

Constructor Summary

Constructors
Constructor and Description

ConfusionMatrix()

Constructors
Constructor and Description
`ConfusionMatrix()`

Method Summary

Methods
Modifier and Type	Method and Description
`static ConfusionMatrix`	`createCumulativeMatrix(ConfusionMatrix... matrices)` Sums up all matrices into a new one
`double`	`getAccuracy()`
`double`	`getAvgPrecision()`
`double`	`getAvgRecall()`
`double`	`getCohensKappa()` Computes Cohen's Kappa
`int`	`getColSum(String label)`
`double`	`getConfidence90Accuracy()` Returns the half of the confidence interval on accuracy on alpha = 90
`double`	`getConfidence90AccuracyHigh()`
`double`	`getConfidence90AccuracyLow()`
`double`	`getConfidence90MacroFM()`
`double`	`getConfidence95Accuracy()` Returns the half of the confidence interval on accuracy on alpha = 95
`double`	`getConfidence95AccuracyHigh()` Returns the upper bound of the accuracy with alpha = 95
`double`	`getConfidence95AccuracyLow()` Returns the lower bound of the accuracy with alpha = 95
`double`	`getConfidence95MacroFM()` Returns the half of confidence interval on alpha = 95 (see http://alias-i.com/lingpipe/docs/api/com/aliasi/classify/ConfusionMatrix.html#confidence95%28%29)
`double`	`getConfidence95MacroFMHigh()` Returns the upper bound of the macro F-measure with alpha = 95
`double`	`getConfidence95MacroFMLow()` Returns the lower bound of the macro F-measure with alpha = 95
`Map<String,Double>`	`getFMeasureForLabels()` Returns F-measure for categories; see See http://en.wikipedia.org/wiki/F1_score
`Map<String,Double>`	`getFMeasureForLabels(double beta)` See http://en.wikipedia.org/wiki/F1_score
`List<String>`	`getLabelSeries()` Returns the series of actual labels
`double`	`getMacroFMeasure()` Macro-averaged F-measure gives equal weight to each category, regardless of its frequency.
`double`	`getMacroFMeasure(double beta)` Macro-averaged F-measure with beta parameter.
`double`	`getMicroFMeasure()` Micro-averaged F-measure gives equal weight to each document and is therefore considered as an average over all the document/category pairs.
`double`	`getPrecisionForLabel(String label)`
`Map<String,Double>`	`getPrecisionForLabels()`
`double`	`getRecallForLabel(String label)` Return recall for single label
`Map<String,Double>`	`getRecallForLabels()` Return recall for labels
`int`	`getRowSum(String label)`
`ConfusionMatrix`	`getSymmetricConfusionMatrix()` Returns C + C^T - InC (this + transposed this - unit matrix * this), see See Silvie Cinkova, Martin Holub, and Vincent Kriz. 2012.
`int`	`getTotalSum()`
`ConfusionMatrix`	`getTransposedMatrix()` Returns transposed confusion matrix (gold and predicted are switched)
`void`	`increaseValue(String goldValue, String observedValue)`
`void`	`increaseValue(String goldValue, String observedValue, int times)` Increases value of goldValue x observedValue n times
`static ConfusionMatrix`	`parseFromText(String text)` Confusion matrix printed to text by toString can be parsed back
`String`	`printClassDistributionGold()` Returns a distribution of classes in gold data and predicted data (absolute and relative)
`String`	`printLabelPrecRecFm()`
`String`	`printNiceResults()`
`void`	`setNumberOfDecimalPlaces(int numberOfDecimalPlaces)`
`protected String`	`tableToString(List<List<String>> table)`
`String`	`toString()`
`String`	`toStringLatex()` Prints in LaTeX format
`String`	`toStringProbabilistic()` Returns probabilistic confusion matrix (as table String)

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

- Constructor Detail
  - ConfusionMatrix
```
public ConfusionMatrix()
```
- Method Detail
  - setNumberOfDecimalPlaces
```
public void setNumberOfDecimalPlaces(int numberOfDecimalPlaces)
                              throws IllegalArgumentException
```
    Throws:
    
    IllegalArgumentException
  - increaseValue
```
public void increaseValue(String goldValue,
                 String observedValue)
```
  - getLabelSeries
```
public List<String> getLabelSeries()
```
    Returns the series of actual labels
    
    Returns:
    list of labels
  - increaseValue
```
public void increaseValue(String goldValue,
                 String observedValue,
                 int times)
```
    Increases value of goldValue x observedValue n times
    
    Parameters:
    goldValue - exp
    observedValue - ac
    times - n-times
  - getAccuracy
```
public double getAccuracy()
```
  - getTotalSum
```
public int getTotalSum()
```
  - getRowSum
```
public int getRowSum(String label)
```
  - getColSum
```
public int getColSum(String label)
```
  - getPrecisionForLabels
```
public Map<String,Double> getPrecisionForLabels()
```
  - getPrecisionForLabel
```
public double getPrecisionForLabel(String label)
```
  - getMicroFMeasure
```
public double getMicroFMeasure()
```
    Micro-averaged F-measure gives equal weight to each document and is therefore considered as an average over all the document/category pairs. It tends to be dominated by the classifierâ€™s performance on common categories. (It's actually the accuracy).
    (from Ozgur et al., 2005. Text Categorization with Class-Based and Corpus-Based Keyword Selection.)
    
    Returns:
    double
  - getMacroFMeasure
```
public double getMacroFMeasure()
```
    Macro-averaged F-measure gives equal weight to each category, regardless of its frequency. It is influenced more by the classifier performance on rare categories.
    
    Returns:
    double
  - getMacroFMeasure
```
public double getMacroFMeasure(double beta)
```
    Macro-averaged F-measure with beta parameter.
    
    Parameters:
    beta - beta parameter
    
    Returns:
    double
  - getFMeasureForLabels
```
public Map<String,Double> getFMeasureForLabels()
```
    Returns F-measure for categories; see See http://en.wikipedia.org/wiki/F1_score
    
    Returns:
    double
  - getFMeasureForLabels
```
public Map<String,Double> getFMeasureForLabels(double beta)
```
    See http://en.wikipedia.org/wiki/F1_score
    
    Parameters:
    beta - beta paremeter; higher than 1 prefers recall, lower than 1 prefers precision
    
    Returns:
    double
  - getRecallForLabels
```
public Map<String,Double> getRecallForLabels()
```
    Return recall for labels
    
    Returns:
    double
  - getRecallForLabel
```
public double getRecallForLabel(String label)
```
    Return recall for single label
    
    Parameters:
    label - label
    
    Returns:
    double
  - getConfidence95Accuracy
```
public double getConfidence95Accuracy()
```
    Returns the half of the confidence interval on accuracy on alpha = 95
    
    Returns:
    conf. int
  - getConfidence90Accuracy
```
public double getConfidence90Accuracy()
```
    Returns the half of the confidence interval on accuracy on alpha = 90
    
    Returns:
    conf. int
  - getConfidence90AccuracyLow
```
public double getConfidence90AccuracyLow()
```
  - getConfidence90AccuracyHigh
```
public double getConfidence90AccuracyHigh()
```
  - getConfidence95AccuracyLow
```
public double getConfidence95AccuracyLow()
```
    Returns the lower bound of the accuracy with alpha = 95
    
    Returns:
    accuracy minus half of the confidence interval
  - getConfidence95AccuracyHigh
```
public double getConfidence95AccuracyHigh()
```
    Returns the upper bound of the accuracy with alpha = 95
    
    Returns:
    accuracy plus half of the confidence interval
  - getConfidence95MacroFM
```
public double getConfidence95MacroFM()
```
    Returns the half of confidence interval on alpha = 95 (see http://alias-i.com/lingpipe/docs/api/com/aliasi/classify/ConfusionMatrix.html#confidence95%28%29)
    
    Returns:
    conf
  - getConfidence90MacroFM
```
public double getConfidence90MacroFM()
```
  - getConfidence95MacroFMLow
```
public double getConfidence95MacroFMLow()
```
    Returns the lower bound of the macro F-measure with alpha = 95
    
    Returns:
    macro F-measure minus half of the confidence interval
  - getConfidence95MacroFMHigh
```
public double getConfidence95MacroFMHigh()
```
    Returns the upper bound of the macro F-measure with alpha = 95
    
    Returns:
    macro F-measure plus half of the confidence interval
  - getCohensKappa
```
public double getCohensKappa()
```
    Computes Cohen's Kappa
    
    Returns:
    double
  - tableToString
```
protected String tableToString(List<List<String>> table)
```
  - toString
```
public String toString()
```
    Overrides:
    
    toString in class Object
  - toStringLatex
```
public String toStringLatex()
```
    Prints in LaTeX format
    
    Returns:
    string
  - printNiceResults
```
public String printNiceResults()
```
  - printLabelPrecRecFm
```
public String printLabelPrecRecFm()
```
  - getAvgPrecision
```
public double getAvgPrecision()
```
  - getAvgRecall
```
public double getAvgRecall()
```
  - createCumulativeMatrix
```
public static ConfusionMatrix createCumulativeMatrix(ConfusionMatrix... matrices)
```
    Sums up all matrices into a new one
    
    Parameters:
    matrices - confusion matrices
    
    Returns:
    confusion matrix
  - getSymmetricConfusionMatrix
```
public ConfusionMatrix getSymmetricConfusionMatrix()
```
    Returns C + C^T - InC (this + transposed this - unit matrix * this), see
    See Silvie Cinkova, Martin Holub, and Vincent Kriz. 2012. Managing uncertainty in semantic tagging. In Proceedings of the 13th Conference of the European Chapter of the Association for Computational Linguistics, EACL '12, pages 840-850, Stroudsburg, PA, USA. Association for Computational Linguistics.
    
    Returns:
    new instance
  - getTransposedMatrix
```
public ConfusionMatrix getTransposedMatrix()
```
    Returns transposed confusion matrix (gold and predicted are switched)
    
    Returns:
    new instance
  - parseFromText
```
public static ConfusionMatrix parseFromText(String text)
                                     throws IllegalArgumentException
```
    Confusion matrix printed to text by toString can be parsed back
    
    Parameters:
    text - input text
    
    Returns:
    confusion matrix
    
    Throws:
    
    IllegalArgumentException - if input is malformed
  - printClassDistributionGold
```
public String printClassDistributionGold()
```
    Returns a distribution of classes in gold data and predicted data (absolute and relative)
    
    Returns:
    string
  - toStringProbabilistic
```
public String toStringProbabilistic()
```
    Returns probabilistic confusion matrix (as table String)
    
    Returns:
    string

Class ConfusionMatrix

Constructor Summary

Method Summary

Methods inherited from class java.lang.Object

Constructor Detail

ConfusionMatrix

Method Detail

setNumberOfDecimalPlaces

increaseValue

getLabelSeries

increaseValue

getAccuracy

getTotalSum

getRowSum

getColSum

getPrecisionForLabels

getPrecisionForLabel

getMicroFMeasure

getMacroFMeasure

getMacroFMeasure

getFMeasureForLabels

getFMeasureForLabels

getRecallForLabels

getRecallForLabel

getConfidence95Accuracy

getConfidence90Accuracy

getConfidence90AccuracyLow

getConfidence90AccuracyHigh

getConfidence95AccuracyLow

getConfidence95AccuracyHigh

getConfidence95MacroFM

getConfidence90MacroFM

getConfidence95MacroFMLow

getConfidence95MacroFMHigh

getCohensKappa

tableToString

toString

toStringLatex

printNiceResults

printLabelPrecRecFm

getAvgPrecision

getAvgRecall

createCumulativeMatrix

getSymmetricConfusionMatrix

getTransposedMatrix

parseFromText

printClassDistributionGold

toStringProbabilistic