Mj Baxter

Model-based cluster analysis of artefact compositional data

Cluster analysis is the most widely used multivariate technique in archaeometry, with the majority of applications being exploratory in nature. Model-based methods of clustering have their advocates, but have seen little application to archaeometric data. The paper investigates two such methods. They have potential advantages over exploratory techniques, if successful. Mixture maximum-likelihood worked well using low-dimensional lead isotope data, but had problems coping with higher-dimensional ceramic compositional data. For our most challenging example, classification maximum-likelihood performed comparably with more standard methods, but we find no evidence to suggest that it should supplant these.

Variable Selection in Artefact Compositional Studies

In artefact compositional studies, the selection of variables to use in analysis is unavoidable. Given this ubiquity, surprisingly little attention has been paid to ways in which variables might be selected. After arguing the case for the importance of variable selection, two systematic approaches to making a choice, which have had little or no application in archaeometry, are discussed and illustrated. One, based on the use of principal components, is appropriate if structure in the data is not known. The other, based on the use of classification trees, is applicable when there are known or assumed groups in the data.