A Dissimilarity matrix can be used in conjunction with Multidimensional scaling to aid visualization of high-dimensional data. A Dissimilarity object is a matrix of the distances, according to the chosen distance function, between all the data points in the PatternList object.
A Dissimilarity object can be created by selecting a PatternList object and choosing To Dissimilarity. The dissimilarity matrix can also be computed using feature weights. This is done by selecting a PatternList object, an FeatureWeights object and choosing To Dissimilarity.
© Ola Söder, May 29, 2008