Multivariate Graphics
Daniel B. Carr · Wiley StatsRef: Statistics Reference Online · 2014
Abstract This extensive tour of multivariate graphics leads into methods that can be applied to biostatistics. Basic design issues and techniques, such as smoothing and density estimation, are considered. Issues of choice of color, depth, and structure are discussed. Multivariate representations basically fall into a few classes: glyph plots, linked plots, nested plots, conditioned plots, geometric section plots, series plots, and composite plots. Each of these classes is described in detail. Once the graphics are created, important considerations of viewing the graphics in terms of dimensionality, cognostics, and projection pursuit arise. Ways of handling these issues by starting from low‐dimensional plots and seeing what happens through cognostics or projection pursuit, and sectioning are pursued. The ongoing challenges of producing graphics from massive datasets is also considered.