Package at.tuwien.ifs.somtoolbox.layers.metrics

This package contains the metrics used for distance calculation during training and for mapping data onto maps.

See:
          Description

Interface Summary
DistanceMetric The interface, distance metric classes have to implement.
 

Class Summary
AbstractMetric Implements a static method for metric instantiation and a method for mean vector calculation.
AbstractWeightedMetric  
CosineMetric Implements the cosine metric, defined for two vectors d1 and d2 as d1xd2 / (|d1|*|d2|).
L1Metric Implements the L1 or city block metric.
L2Metric Implements the L2 or Euclidean metric.
L2MetricFast Implements a fast version of the L2 or Euclidean metric, by not taking the square root.
L2MetricSparse Implements the L2 or Euclidean metric, considering only those values for the distance calculation that have non-zero values for the first, second or both vectors, depending on the initialisation mode.
L2MetricWeighted  
LInfinityMetric Implements the L-Infinity metric,defined for two vectors x and y as max( |xi-yi| ), i = 1,...,|x|.
LnAlphaMetric  
LnMetric Generic Ln metric.
MahalanobisMetric Implements the Mahalanobis distance metric.
Metrics  
MissingValueMetricWrapper A wrapper class around other distance metrics, modifying the distance computation in such a way that only vector attributes that are not missing (indicated by InputData.MISSING_VALUE are considered.
When instantiating using the empty constructor MissingValueMetricWrapper.MissingValueMetricWrapper() the default metric MissingValueMetricWrapper.DEFAULT_METRIC is used.
MnemonicSOMMetric A metric for mnemonic SOMs.
 

Enum Summary
DistanceMetric.SparcseMetricModes  
 

Exception Summary
MetricException Is thrown if vectors with different dimensionalities are subject to mathematical operations.
 

Package at.tuwien.ifs.somtoolbox.layers.metrics Description

This package contains the metrics used for distance calculation during training and for mapping data onto maps.