GRAPHICAL REPRESENTATIONS AS A DISCIPLINE

Abstract : In recent years, partly as a result of the computer revolution, the use of graphical representations in Statistics is beginning to attract more attention. It seems appropriate to develop a discipline for such graphical representations. It is proposed that one way to develop such a discipline is to compile a list of functions and attributes on which each graphical method would be scaled and on which each potential application would also be scaled. For each application, the appropriate method could be selected by how well it matches the application in these attribute scales. Several representations of multivariate data including profiles, stars, faces and Andrews' Fourier Series are used as illustrations. (Author)