A paper about k-means analysis with outlier detection, mixed variable types and missing values was presented at the German Classification Society in March 2001. A technical critique of k-means clustering ishere and a pre-publication copy of the GfKl 2001 paper can be downloaded as a zip file (65Kb).
Our revised classification of single malt whiskies was presented at the International Federation of Classification Societies conference in July 2000, and at the British Computer Society and Royal Statistical Society in December 2000. The classification is availablehere.
Please also refer to a newTwo-Stage k-Means clustering method we have recently developed, and which was presented at GfKl 2000. This method is described in the FocalPoint Clustering User Guide , which is supplied with ClustanGraphics5.
Our methodology for large-scale clustering was presented at the 1999 meeting of the International Statistical Institute in Helsinki. A 4-page Abstract was published in the ISI '99 proceedings and can bedownloaded here as a zip file (100 Kb).
At the 1998 meeting of the German Classification Society in Dresden we discussed optimal seriation of a hierarchical classification, and ClustanGraphics. The paper was subsequently published in Classification in the Information Age, Gaul, W. and Locarek-Junge, H. (Eds), Springer 1999, pp 268-275. The paper can bedownloaded as a zip file, by kind permission of Springer-Verlag (70 Kb).
A paper was presented at Interface 1998 in Minneapolis, USA. It describes our unique methodology for large-scale clustering, and was subsequently published in Computing Science and Statistics, 30, pp257-263. The paper can bedownloaded as a zip file (65Kb).