‘Size’ and ‘Shape’ in the Measurement of Multivariate Proximity

Ordination and clustering methods are widely applied to ecological data that are non-negative (like species abundances or biomasses). These methods rely on a measure of multivariate proximity that quantifies differences between the sampling units (e.g. individuals, stations, time points), leading to results such as:

  1. Ordinations of the units, where interpoint distances optimally display the measured differences
  2. Clustering the units into homogeneous clusters
  3. Assessing differences between pre-specified groups of units (e.g. regions, periods, treatment–control groups)

In this video, Michael Greenacre introduces his new article, ‘‘Size’ and ‘Shape’ in the Measurement of Multivariate Proximity’, published in Methods in Ecology and Evolution, May 2017. In the context of species abundances, for example, he explains how much a chosen proximity measure captures the difference in “size” between two samples, i.e. difference in overall abundances, and differences in “shape”, i.e. differences in compositions or relative abundances.  He shows that the popular Bray-Curtis dissimilarity inevitably includes a part of the “size” difference in its measurement of multivariate proximity.

This video is based on the article ‘‘Size’ and ‘shape’ in the measurement of multivariate proximity‘ by Michael Greenacre.

Advertisements