ConfusionMatrix (h2o-core 3.0.0 API)

java.lang.Object
- water.Iced
- - hex.ConfusionMatrix

All Implemented Interfaces:

java.io.Externalizable, java.io.Serializable, java.lang.Cloneable, Freezable
```
public class ConfusionMatrix
extends Iced
```
See Also:
Serialized Form

Field Summary

Fields
Modifier and Type Field and Description

double[][] _cm

java.lang.String[] _domain

Fields
Modifier and Type	Field and Description
`double[][]`	`_cm`
`java.lang.String[]`	`_domain`

Constructor Summary

Constructors
Constructor and Description

ConfusionMatrix(double[][] value, java.lang.String[] domain)
Constructor for Confusion Matrix

Constructors
Constructor and Description
`ConfusionMatrix(double[][] value, java.lang.String[] domain)` Constructor for Confusion Matrix

Method Summary

Methods
Modifier and Type	Method and Description
`double`	`accuracy()` The percentage of predictions that are correct.
`void`	`add(ConfusionMatrix other)`
`void`	`add(int i, int j)`
`static ConfusionMatrix`	`buildCM(Vec actuals, Vec predictions)` Build the CM data from the actuals and predictions, using the default threshold.
`double`	`classErr(int c)`
`double`	`err()`
`long`	`errCount()`
`double`	`F0point5()` Returns the F-measure which combines precision and recall and weights precision higher than recall.
`double`	`F1()` Returns the F-measure which combines precision and recall.
`double`	`F2()` Returns the F-measure which combines precision and recall and weights recall higher than precision.
`boolean`	`isBinary()`
`double`	`max_per_class_error()` The maximum per-class error
`double`	`mcc()` The Matthews Correlation Coefficient, takes true negatives into account in contrast to F-Score See MCC MCC = Correlation between observed and predicted binary classification
`int`	`nclasses()`
`double`	`precision()` The percentage of positive predictions that are correct.
`double`	`recall()` The percentage of positive labeled instances that were predicted as positive.
`int`	`size()`
`double`	`specificity()` The percentage of negative labeled instances that were predicted as negative.
`TwoDimTable`	`table()` Convert this ConfusionMatrix into a fully annotated TwoDimTable
`java.lang.String`	`toASCII()`
`java.lang.String`	`toString()`
`long`	`totalRows()`

Methods inherited from class water.Iced
clone, frozenType, read_impl, read, readExternal, readJSON_impl, readJSON, toJsonString, write_impl, write, writeExternal, writeHTML_impl, writeHTML, writeJSON_impl, writeJSON

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

- Field Detail
  - _cm
```
public final double[][] _cm
```
  - _domain
```
public final java.lang.String[] _domain
```
- Constructor Detail
  - ConfusionMatrix
```
public ConfusionMatrix(double[][] value,
               java.lang.String[] domain)
```
    Constructor for Confusion Matrix
    
    Parameters:
    value - 2D square matrix with co-occurrence counts for actual vs predicted class membership
    domain - class labels (unified domain between actual and predicted class labels)
- Method Detail
  - buildCM
```
public static ConfusionMatrix buildCM(Vec actuals,
                      Vec predictions)
```
    Build the CM data from the actuals and predictions, using the default threshold. Print to Log.info if the number of classes is below the print_threshold. Actuals might have extra levels not trained on (hence never predicted). Actuals with NAs are not scored, and their predictions ignored.
  - add
```
public void add(int i,
       int j)
```
  - size
```
public final int size()
```
  - classErr
```
public final double classErr(int c)
```
  - totalRows
```
public long totalRows()
```
  - add
```
public void add(ConfusionMatrix other)
```
  - err
```
public double err()
```
    Returns:
    overall classification error
  - errCount
```
public long errCount()
```
  - accuracy
```
public double accuracy()
```
    The percentage of predictions that are correct.
  - specificity
```
public double specificity()
```
    The percentage of negative labeled instances that were predicted as negative.
    
    Returns:
    TNR / Specificity
  - recall
```
public double recall()
```
    The percentage of positive labeled instances that were predicted as positive.
    
    Returns:
    Recall / TPR / Sensitivity
  - precision
```
public double precision()
```
    The percentage of positive predictions that are correct.
    
    Returns:
    Precision
  - mcc
```
public double mcc()
```
    The Matthews Correlation Coefficient, takes true negatives into account in contrast to F-Score See MCC MCC = Correlation between observed and predicted binary classification
    
    Returns:
    mcc ranges from -1 (total disagreement) ... 0 (no better than random) ... 1 (perfect)
  - max_per_class_error
```
public double max_per_class_error()
```
    The maximum per-class error
    
    Returns:
    max[classErr(i)]
  - nclasses
```
public final int nclasses()
```
  - isBinary
```
public final boolean isBinary()
```
  - F1
```
public double F1()
```
    Returns the F-measure which combines precision and recall.
    C.f. end of http://en.wikipedia.org/wiki/Precision_and_recall.
  - F2
```
public double F2()
```
    Returns the F-measure which combines precision and recall and weights recall higher than precision.
    See F1_score
  - F0point5
```
public double F0point5()
```
    Returns the F-measure which combines precision and recall and weights precision higher than recall.
    See F1_score
  - toString
```
public java.lang.String toString()
```
    Overrides:
    
    toString in class java.lang.Object
  - toASCII
```
public java.lang.String toASCII()
```
  - table
```
public TwoDimTable table()
```
    Convert this ConfusionMatrix into a fully annotated TwoDimTable
    
    Returns:
    TwoDimTable

Class ConfusionMatrix

Field Summary

Constructor Summary

Method Summary

Methods inherited from class water.Iced

Methods inherited from class java.lang.Object

Field Detail

_cm

_domain

Constructor Detail

ConfusionMatrix

Method Detail

buildCM

add

size

classErr

totalRows

add

err

errCount

accuracy

specificity

recall

precision

mcc

max_per_class_error

nclasses

isBinary

F1

F2

F0point5

toString

toASCII

table