EvaluationBinary (deeplearning4j-nn 1.0.0-beta API)

java.lang.Object
- org.deeplearning4j.eval.BaseEvaluation<EvaluationBinary>
- - org.deeplearning4j.eval.EvaluationBinary

All Implemented Interfaces:

Serializable, IEvaluation<EvaluationBinary>
```
public class EvaluationBinary
extends BaseEvaluation<EvaluationBinary>
```
EvaluationBinary: used for evaluating networks with binary classification outputs. The typical classification metrics, such as accuracy, precision, recall, F1 score, etc. are calculated for each output.
Note that ROCBinary is also used internally to calculate AUC for each output, but only when using an appropriate constructor, EvaluationBinary(int, Integer)
Note that EvaluationBinary supports both per-example and per-output masking.
EvaluationBinary by default uses a decision threshold of 0.5, however decision thresholds can be set on a per-output basis using EvaluationBinary(INDArray).
The most common use case: multi-task networks, where each output is a binary value. This differs from Evaluation in that Evaluation is for a single class (binary or non-binary) evaluation.

Author:

Alex Black

See Also:

Serialized Form

Field Summary

Fields
Modifier and Type Field and Description

static double DEFAULT_EDGE_VALUE

static int DEFAULT_PRECISION

Fields
Modifier and Type	Field and Description
`static double`	`DEFAULT_EDGE_VALUE`
`static int`	`DEFAULT_PRECISION`

Constructor Summary

Constructors
Constructor and Description
`EvaluationBinary(org.nd4j.linalg.api.ndarray.INDArray decisionThreshold)` Create an EvaulationBinary instance with an optional decision threshold array.
`EvaluationBinary(int size, Integer rocBinarySteps)` This constructor allows for ROC to be calculated in addition to the standard evaluation metrics, when the rocBinarySteps arg is non-null.

Method Summary

All Methods Static Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`double`	`accuracy(int outputNum)` Get the accuracy for the specified output
`double`	`averageAccuracy()`
`double`	`averageF1()`
`double`	`averagePrecision()`
`double`	`averageRecall()`
`void`	`eval(org.nd4j.linalg.api.ndarray.INDArray labels, org.nd4j.linalg.api.ndarray.INDArray networkPredictions)`
`void`	`eval(org.nd4j.linalg.api.ndarray.INDArray labels, org.nd4j.linalg.api.ndarray.INDArray networkPredictions, org.nd4j.linalg.api.ndarray.INDArray maskArray)`
`void`	`evalTimeSeries(org.nd4j.linalg.api.ndarray.INDArray labels, org.nd4j.linalg.api.ndarray.INDArray predictions, org.nd4j.linalg.api.ndarray.INDArray labelsMask)`
`double`	`f1(int outputNum)` Get the F1 score for the specified output
`double`	`falseNegativeRate(Integer classLabel)` Returns the false negative rate for a given label
`double`	`falseNegativeRate(Integer classLabel, double edgeCase)` Returns the false negative rate for a given label
`int`	`falseNegatives(int outputNum)` Get the false negatives count for the specified output
`double`	`falsePositiveRate(int classLabel)` Returns the false positive rate for a given label
`double`	`falsePositiveRate(int classLabel, double edgeCase)` Returns the false positive rate for a given label
`int`	`falsePositives(int outputNum)` Get the false positives count for the specified output
`double`	`fBeta(double beta, int outputNum)` Calculate the F-beta value for the given output
`static EvaluationBinary`	`fromJson(String json)`
`static EvaluationBinary`	`fromYaml(String yaml)`
`ROCBinary`	`getROCBinary()` Returns the `ROCBinary` instance, if present
`double`	`gMeasure(int output)` Calculate the G-measure for the given output
`double`	`matthewsCorrelation(int outputNum)` Calculate the Matthews correlation coefficient for the specified output
`void`	`merge(EvaluationBinary other)`
`int`	`numLabels()` Returns the number of labels - (i.e., size of the prediction/labels arrays) - if known.
`double`	`precision(int outputNum)` Get the precision (tp / (tp + fp)) for the specified output
`double`	`recall(int outputNum)` Get the recall (tp / (tp + fn)) for the specified output
`void`	`reset()`
`void`	`setLabelNames(List<String> labels)` Set the label names, for printing via `stats()`
`String`	`stats()` Get a String representation of the EvaluationBinary class, using the default precision
`String`	`stats(int printPrecision)` Get a String representation of the EvaluationBinary class, using the specified precision
`int`	`totalCount(int outputNum)` Get the total number of values for the specified column, accounting for any masking
`int`	`trueNegatives(int outputNum)` Get the true negatives count for the specified output
`int`	`truePositives(int outputNum)` Get the true positives count for the specified output

Methods inherited from class org.deeplearning4j.eval.BaseEvaluation
eval, evalTimeSeries, fromJson, fromYaml, toJson, toString, toYaml

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

- Field Detail
  - DEFAULT_PRECISION
```
public static final int DEFAULT_PRECISION
```
    See Also:
    
    Constant Field Values
  - DEFAULT_EDGE_VALUE
```
public static final double DEFAULT_EDGE_VALUE
```
    See Also:
    
    Constant Field Values
- Constructor Detail
  - EvaluationBinary
```
public EvaluationBinary(org.nd4j.linalg.api.ndarray.INDArray decisionThreshold)
```
    Create an EvaulationBinary instance with an optional decision threshold array.
    
    Parameters:
    
    decisionThreshold - Decision threshold for each output; may be null. Should be a row vector with length equal to the number of outputs, with values in range 0 to 1. An array of 0.5 values is equivalent to the default (no manually specified decision threshold).
  - EvaluationBinary
```
public EvaluationBinary(int size,
                        Integer rocBinarySteps)
```
    This constructor allows for ROC to be calculated in addition to the standard evaluation metrics, when the rocBinarySteps arg is non-null. See ROCBinary for more details
    
    Parameters:
    
    size - Number of outputs
    
    rocBinarySteps - Constructor arg for ROCBinary.ROCBinary(int)
- Method Detail
  - eval
```
public void eval(org.nd4j.linalg.api.ndarray.INDArray labels,
                 org.nd4j.linalg.api.ndarray.INDArray networkPredictions)
```
  - evalTimeSeries
```
public void evalTimeSeries(org.nd4j.linalg.api.ndarray.INDArray labels,
                           org.nd4j.linalg.api.ndarray.INDArray predictions,
                           org.nd4j.linalg.api.ndarray.INDArray labelsMask)
```
    Specified by:
    
    evalTimeSeries in interface IEvaluation<EvaluationBinary>
    
    Overrides:
    
    evalTimeSeries in class BaseEvaluation<EvaluationBinary>
  - eval
```
public void eval(org.nd4j.linalg.api.ndarray.INDArray labels,
                 org.nd4j.linalg.api.ndarray.INDArray networkPredictions,
                 org.nd4j.linalg.api.ndarray.INDArray maskArray)
```
    Specified by:
    
    eval in interface IEvaluation<EvaluationBinary>
    
    Overrides:
    
    eval in class BaseEvaluation<EvaluationBinary>
  - merge
```
public void merge(EvaluationBinary other)
```
  - reset
```
public void reset()
```
  - numLabels
```
public int numLabels()
```
    Returns the number of labels - (i.e., size of the prediction/labels arrays) - if known. Returns -1 otherwise
  - setLabelNames
```
public void setLabelNames(List<String> labels)
```
    Set the label names, for printing via stats()
  - totalCount
```
public int totalCount(int outputNum)
```
    Get the total number of values for the specified column, accounting for any masking
  - truePositives
```
public int truePositives(int outputNum)
```
    Get the true positives count for the specified output
  - trueNegatives
```
public int trueNegatives(int outputNum)
```
    Get the true negatives count for the specified output
  - falsePositives
```
public int falsePositives(int outputNum)
```
    Get the false positives count for the specified output
  - falseNegatives
```
public int falseNegatives(int outputNum)
```
    Get the false negatives count for the specified output
  - averageAccuracy
```
public double averageAccuracy()
```
  - accuracy
```
public double accuracy(int outputNum)
```
    Get the accuracy for the specified output
  - averagePrecision
```
public double averagePrecision()
```
  - precision
```
public double precision(int outputNum)
```
    Get the precision (tp / (tp + fp)) for the specified output
  - averageRecall
```
public double averageRecall()
```
  - recall
```
public double recall(int outputNum)
```
    Get the recall (tp / (tp + fn)) for the specified output
  - averageF1
```
public double averageF1()
```
  - fBeta
```
public double fBeta(double beta,
                    int outputNum)
```
    Calculate the F-beta value for the given output
    
    Parameters:
    
    beta - Beta value to use
    
    outputNum - Output number
    
    Returns:
    
    F-beta for the given output
  - f1
```
public double f1(int outputNum)
```
    Get the F1 score for the specified output
  - matthewsCorrelation
```
public double matthewsCorrelation(int outputNum)
```
    Calculate the Matthews correlation coefficient for the specified output
    
    Parameters:
    
    outputNum - Output number
    
    Returns:
    
    Matthews correlation coefficient
  - gMeasure
```
public double gMeasure(int output)
```
    Calculate the G-measure for the given output
    
    Parameters:
    
    output - The specified output
    
    Returns:
    
    The G-measure for the specified output
  - falsePositiveRate
```
public double falsePositiveRate(int classLabel)
```
    Returns the false positive rate for a given label
    
    Parameters:
    
    classLabel - the label
    
    Returns:
    
    fpr as a double
  - falsePositiveRate
```
public double falsePositiveRate(int classLabel,
                                double edgeCase)
```
    Returns the false positive rate for a given label
    
    Parameters:
    
    classLabel - the label
    
    edgeCase - What to output in case of 0/0
    
    Returns:
    
    fpr as a double
  - falseNegativeRate
```
public double falseNegativeRate(Integer classLabel)
```
    Returns the false negative rate for a given label
    
    Parameters:
    
    classLabel - the label
    
    Returns:
    
    fnr as a double
  - falseNegativeRate
```
public double falseNegativeRate(Integer classLabel,
                                double edgeCase)
```
    Returns the false negative rate for a given label
    
    Parameters:
    
    classLabel - the label
    
    edgeCase - What to output in case of 0/0
    
    Returns:
    
    fnr as a double
  - getROCBinary
```
public ROCBinary getROCBinary()
```
    Returns the ROCBinary instance, if present
  - stats
```
public String stats()
```
    Get a String representation of the EvaluationBinary class, using the default precision
    
    Returns:
  - stats
```
public String stats(int printPrecision)
```
    Get a String representation of the EvaluationBinary class, using the specified precision
    
    Parameters:
    
    printPrecision - The precision (number of decimal places) for the accuracy, f1, etc.
  - fromJson
```
public static EvaluationBinary fromJson(String json)
```
  - fromYaml
```
public static EvaluationBinary fromYaml(String yaml)
```

Class EvaluationBinary

Field Summary

Constructor Summary

Method Summary

Methods inherited from class org.deeplearning4j.eval.BaseEvaluation

Methods inherited from class java.lang.Object

Field Detail

DEFAULT_PRECISION

DEFAULT_EDGE_VALUE

Constructor Detail

EvaluationBinary

EvaluationBinary

Method Detail

eval

evalTimeSeries

eval

merge

reset

numLabels

setLabelNames

totalCount

truePositives

trueNegatives

falsePositives

falseNegatives

averageAccuracy

accuracy

averagePrecision

precision

averageRecall

recall

averageF1

fBeta

f1

matthewsCorrelation

gMeasure

falsePositiveRate

falsePositiveRate

falseNegativeRate

falseNegativeRate

getROCBinary

stats

stats

fromJson

fromYaml