public class DominanceRelationCalculator
extends java.lang.Object
Constructor | Description |
---|---|
DominanceRelationCalculator() |
Sole constructor
|
Modifier and Type | Method | Description |
---|---|---|
java.util.HashSet<PairOfIndices> |
calculateDominanceRelation(MemoryContainer memoryContainer) |
Gets dominance relation, i.e., set of pairs of examples (a,b), where a, b belong to given data set, such that a D b.
|
java.util.HashSet<PairOfIndices> |
calculateDominanceRelation(MemoryContainer memoryContainer,
byte[] attributesMask) |
public DominanceRelationCalculator()
public java.util.HashSet<PairOfIndices> calculateDominanceRelation(MemoryContainer memoryContainer)
memoryContainer
- considered data setpublic java.util.HashSet<PairOfIndices> calculateDominanceRelation(MemoryContainer memoryContainer, byte[] attributesMask)
memoryContainer
- considered data setattributesMask
- Mask for attributes, allowing dominance cone calculations for a subset of active and condition attributes.
It is checked if length of the mask is equal to the number of attributes in memory container.
If attributesMask(i) is equal to zero,
than corresponding attribute is not taken into account at all when calculating dominance cone, even if it's active and condition.
If attributesMask(i) is different than zero,
than corresponding attribute is considered further - it is checked if it's active and condition.
In fact only mask values corresponding to active and condition attributes are important. All other values are
not taken into account and may be any.
If all active and condition attributes are masked (i.e., all mask entries corresponding to active and
condition attributes are equal to zero), then resulting dominance cone will contain all examples from given memory container.