Three-dimensional data display using kernel density estimates
CC Beardah, M. J. Baxter · Nottingham Trent University's Institutional Repository (Nottingham Trent Repository) · 1999
Having looked at univariate and bivariate Kernel Density Estimates (KDEs) and their applications in previous CAA proceedings (Beardah and Baxter, 1996a and Beardah, 1998), it is natural to ask how three-dimensional KDEs may be utilised in archaeology. Data that are naturally threedimensional do exist in archaeological applications; for example, we shall later look at the case of lead isotope ratio data. However, the utility of these methods is not restricted to such special cases. Often, higher dimensional data are analysed, by subjecting them to some dimension reduction technique, such as principal component analysis (PCA). KDE methods can then be applied to the first two or three components of the PCA scores, in an effort to identify structure.