HierarchicalClusteringSimilarityDistanceMeasure Enumeration TIBCO Spotfire 6.5 API Reference
Represents a Similarity or a Distance Measure that can be used in Hierarchical Clustering.

Namespace: Spotfire.Dxp.Data.Computations.Clustering
Assembly: Spotfire.Dxp.Data (in Spotfire.Dxp.Data.dll) Version: 14.10.7525.5058 (14.10.7525.5058)
Syntax

[SerializableAttribute]
[PersistenceVersionAttribute(6, 0)]
public enum HierarchicalClusteringSimilarityDistanceMeasure
Members

  Member nameValueDescription
Correlation1 Correlation, Pearson's correlation, Pearson's r.

Subtract the mean value of the coordinates, normalize and take the scalar product.

UnsignedCorrelation2 Unsigned correlation is absolut value of correlation.
CosineCorrelation3 Normalized scalar product without subtracting the mean value.
Tanimoto4 Tanimoto Coefficient, normally only used for binary valued data.
Euclidean5 The ordinary distance in n-dimensional space.
CityBlock6 City block distance, Manhattan distance, L1 distance, rectilinear distance.
SquareEuclidean7 The Euclidean distance squared.
HalfSquareEuclidean8 The Euclidean distance squared and divided by 2; mandatory for Ward's clustering method.
Remarks

Similarities with ranges: Correlation [1, -1], UnsignedCorrelation [1, 0], CosineCorrelation [1, -1], Tanimoto [1, -1/3] ([1, 0] on binary input). Similarities has 1 as most similar value and lower value for less similar.

Distances all have range [0, float.MaxValue]: Euclidean, CityBlock, SquaredEuclidean, HalfSquaredEuclidean. Distances has 0 as smallest distance and higher value for more distant pairs of objects.

Version Information

Supported in: 6.5, 6.0, 5.5, 5.0, 4.5
See Also