ClassificationStatistics (magpie API)

java.lang.Object
- magpie.statistics.performance.BaseStatistics
- - magpie.statistics.performance.ClassificationStatistics

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable, Commandable, Options, Printable, Savable
```
public class ClassificationStatistics
extends BaseStatistics
```
Computes performance statistics for classifiers. Notes about the classification statistics:
- The 'True' class is class #0
- For cases with >2 classes, performance statistics for binary classification (e.g., True Positive Rate) are computed for the simplified problem of 'class 0' vs 'not class 0'
See Also:

Serialized Form

Field Summary

Fields
Modifier and Type	Field and Description
`double`	`Accuracy` Accuracy
`double`	`BaselineLogLoss` Log-loss of zero-r classifier.
`java.lang.String[]`	`ClassNames` Names of classes (used when printing)
`int[][]`	`ConfusionMatrix` Contingency table.
`int[][]`	`ContingencyTable` Description of how often instances are assigned to each class.
`double`	`F1` F1 Score
`double`	`FDR` False Detection Rate
`int`	`FN` Number of False Negatives
`int`	`FP` Number of False Positives
`double`	`FPR` False Positive Rate
`double`	`FractionCorrect` Fraction of entries assigned to correct class
`double`	`Kappa` Fraction correct modified such that 0 is random guessing and 1 is perfect classification
`double`	`LogLoss` Log-loss
`double`	`MCC` Matthews Correlation Coefficient
`double`	`MeasuredLargestClassFraction` Fraction of entries in the most-prevalent class
`double`	`NPV` Negative Predictive Value
`int`	`NumberCorrect` Number of entries assigned to the correct class
`double`	`PPV` Positive Predictive Value
`double`	`Sensitivity` Sensitivity (aka True Positive Rate
`double`	`Specificity` Specificity (True Negative Rate)
`int`	`TN` Number of True Negatives
`int`	`TP` Number of True Positives.

Fields inherited from class magpie.statistics.performance.BaseStatistics
Measured, NumberTested, Predicted, ROC, ROC_AUC

Constructor Summary

Constructors
Constructor and Description

ClassificationStatistics()

Constructors
Constructor and Description
`ClassificationStatistics()`

Method Summary

All Methods Static Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`java.lang.Object`	`clone()`
`protected static double`	`computeLogLoss(int[] trueLabels, double[][] probabilites)` Compute the log-loss for the classifier.
`static int[]`	`discretize(double[] x, int n_classes)` Takes a continuous class variable and converts it to discrete values that range between [0,nclass).
`protected void`	`evaluate_protected(Dataset data)` Internal method for acutally computing results.
`java.util.Map<java.lang.String,java.lang.Double>`	`getStatistics()` Retrieve list of statistics stored in this class.
`java.lang.String`	`printBaselineStats()` Print statistics about the measured class values.
`java.lang.String`	`printCommand(java.util.List<java.lang.String> Command)` Handles more complicated printing commands.
`java.lang.String`	`printContingencyTable()` Print out the contingency table
`java.lang.String`	`toString()`

Methods inherited from class magpie.statistics.performance.BaseStatistics
about, evaluate, getROCCurve, getStatisticsNoNaNs, printDescription, printROCCurve, printUsage, runCommand, saveCommand, savePerformanceData, setOptions

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

- Field Detail
  - NumberCorrect
```
public int NumberCorrect
```
    Number of entries assigned to the correct class
  - FractionCorrect
```
public double FractionCorrect
```
    Fraction of entries assigned to correct class
  - Kappa
```
public double Kappa
```
    Fraction correct modified such that 0 is random guessing and 1 is perfect classification
  - ContingencyTable
```
public int[][] ContingencyTable
```
    Description of how often instances are assigned to each class. Same as ContingencyTable for binary cases. Entry [i][j] is the number of entries classified as class j that were measured to be class i.
  - ConfusionMatrix
```
public int[][] ConfusionMatrix
```
    Contingency table. For binary cases, this is equivalent to ConfusionMatrix. For cases with >2 classes, this is the confusion matrix for a simplified classification problem of "not class 0" vs "class 0."
  - TP
```
public int TP
```
    Number of True Positives. For >2 classes, the this the true positive rate of correctly identifying the first class
  - FP
```
public int FP
```
    Number of False Positives
  - FN
```
public int FN
```
    Number of False Negatives
  - TN
```
public int TN
```
    Number of True Negatives
  - Sensitivity
```
public double Sensitivity
```
    Sensitivity (aka True Positive Rate
  - FPR
```
public double FPR
```
    False Positive Rate
  - Accuracy
```
public double Accuracy
```
    Accuracy
  - Specificity
```
public double Specificity
```
    Specificity (True Negative Rate)
  - PPV
```
public double PPV
```
    Positive Predictive Value
  - NPV
```
public double NPV
```
    Negative Predictive Value
  - FDR
```
public double FDR
```
    False Detection Rate
  - MCC
```
public double MCC
```
    Matthews Correlation Coefficient
  - F1
```
public double F1
```
    F1 Score
  - LogLoss
```
public double LogLoss
```
    Log-loss
  - ClassNames
```
public java.lang.String[] ClassNames
```
    Names of classes (used when printing)
  - MeasuredLargestClassFraction
```
public double MeasuredLargestClassFraction
```
    Fraction of entries in the most-prevalent class
  - BaselineLogLoss
```
public double BaselineLogLoss
```
    Log-loss of zero-r classifier. This is the log-loss of a model that predicts the fractions of each class in the training set as the class probabilities for each prediction.
- Constructor Detail
  - ClassificationStatistics
```
public ClassificationStatistics()
```
- Method Detail
  - discretize
```
public static int[] discretize(double[] x,
                               int n_classes)
```
    Takes a continuous class variable and converts it to discrete values that range between [0,nclass).
    
    Parameters:
    
    x - Array of continuous class variables to be adjusted
    
    n_classes - Number of discrete classes known
    
    Returns:
    
    Class variables adjusted to exist only on integer values (see above)
  - computeLogLoss
```
protected static double computeLogLoss(int[] trueLabels,
                                       double[][] probabilites)
```
    Compute the log-loss for the classifier.
    
    Parameters:
    
    trueLabels - True labels for these predictions
    
    probabilites - Probability of membership in each class
    
    Returns:
    
    Mean log loss over all entries
  - evaluate_protected
```
protected void evaluate_protected(Dataset data)
```
    Description copied from class: BaseStatistics
    
    Internal method for acutally computing results. BaseStatistics.evaluate(Dataset) contains some error-checking code
    
    Specified by:
    
    evaluate_protected in class BaseStatistics
    
    Parameters:
    
    data - Dataset to be evaluated
  - clone
```
public java.lang.Object clone()
                       throws java.lang.CloneNotSupportedException
```
    Overrides:
    
    clone in class BaseStatistics
    
    Throws:
    
    java.lang.CloneNotSupportedException
  - printBaselineStats
```
public java.lang.String printBaselineStats()
```
    Description copied from class: BaseStatistics
    
    Print statistics about the measured class values. Useful as context for interpreting how good the performance statistics actually are
    
    Specified by:
    
    printBaselineStats in class BaseStatistics
    
    Returns:
    
    String listing all of the measured baseline statistics
  - toString
```
public java.lang.String toString()
```
    Overrides:
    
    toString in class java.lang.Object
  - getStatistics
```
public java.util.Map<java.lang.String,java.lang.Double> getStatistics()
```
    Description copied from class: BaseStatistics
    
    Retrieve list of statistics stored in this class.
    
    Specified by:
    
    getStatistics in class BaseStatistics
    
    Returns:
    
    Map of statistic name to value.
  - printContingencyTable
```
public java.lang.String printContingencyTable()
```
    Print out the contingency table
    
    Returns:
    
    Formatted contingency table
  - printCommand
```
public java.lang.String printCommand(java.util.List<java.lang.String> Command)
                              throws java.lang.Exception
```
    Description copied from interface: Printable
    
    Handles more complicated printing commands.
    
    Specified by:
    
    printCommand in interface Printable
    
    Overrides:
    
    printCommand in class BaseStatistics
    
    Parameters:
    
    Command - Command specifying what to print
    
    Returns:
    
    String formatted as requested
    
    Throws:
    
    java.lang.Exception - If command not understood

Class ClassificationStatistics

Field Summary

Fields inherited from class magpie.statistics.performance.BaseStatistics

Constructor Summary

Method Summary

Methods inherited from class magpie.statistics.performance.BaseStatistics

Methods inherited from class java.lang.Object

Field Detail

NumberCorrect

FractionCorrect

Kappa

ContingencyTable

ConfusionMatrix

TP

FP

FN

TN

Sensitivity

FPR

Accuracy

Specificity

PPV

NPV

FDR

MCC

F1

LogLoss

ClassNames

MeasuredLargestClassFraction

BaselineLogLoss

Constructor Detail

ClassificationStatistics

Method Detail

discretize

computeLogLoss

evaluate_protected

clone

printBaselineStats

toString

getStatistics

printContingencyTable

printCommand