RecPair.ConfScoreComparator (ibi Patterns Learn Library Documentation)

java.lang.Object
- com.tibco.patterns.learn.api.project.RecPair.ConfScoreComparator

All Implemented Interfaces:

java.io.Serializable, java.util.Comparator<RecPair>

Enclosing class:

RecPair
```
public static final class RecPair.ConfScoreComparator
extends java.lang.Object
implements java.util.Comparator<RecPair>, java.io.Serializable
```
Provides ordering by a combination (weighted sum) of the prediction confidence (ascending) and the distance of the model score from the 0.5 threshold (ascending). The proportion (importance) of the two parts can be customized. The comparator is consistent with equals: pairs that have the same combination of confidence and score, or no model prediction are ordered by record keys.

See Also:

Serialized Form

- Constructor Summary
  
  Constructors
  Constructor Description
  
  ConfScoreComparator(double confProportion, double scoreDistFrom)
  Creates a ConfScoreComparator with given parameters.
- Method Summary
  
  All Methods Instance Methods Concrete Methods
  Modifier and Type Method Description
  
  int compare(RecPair pair1, RecPair pair2)
  - Methods inherited from class java.lang.Object
    clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
  - Methods inherited from interface java.util.Comparator
    equals, reversed, thenComparing, thenComparing, thenComparing, thenComparingDouble, thenComparingInt, thenComparingLong

- Constructor Detail
  - ConfScoreComparator
```
public ConfScoreComparator(double confProportion,
                           double scoreDistFrom)
```
    Creates a ConfScoreComparator with given parameters.
    
    Parameters:
    
    confProportion - - the proportion (importance) of confidence for the overall ordering (the remaining proportion is model score distance from 0.5). If 1, sorting is by confidence only. If 0, sorting is by score distance only.
    
    scoreDistFrom - - the point from which the model score distance is calculated. If 0, the unmodified score is used. If RLinkOut.RLINK_THRESHOLD, the distance is a measure of how certain the model prediction is. Ignored if confProportion is 1.
    
    Throws:
    
    java.lang.IllegalArgumentException - if any parameter is not in [0; 1].
- Method Detail
  - compare
```
public int compare(RecPair pair1,
                   RecPair pair2)
```
    Specified by:
    
    compare in interface java.util.Comparator<RecPair>