Trellis Displays vs. interactive graphics (Q1965977)

The paper is devoted to comparison of two methods of visualizing high dimensional data: Trellis Displays and interactive statistical graphics. The possibilities and disadvantages of both methods are discussed. The problem is that the number of variables which can be displayed simultaneously in statistical graphics is very limited. Trellis Displays solves the problem in such a way. The variables of the dataset are assigned to the three groups: axis, conditioning and adjunct variables. A specific plot in axis variables is drawn for different subsets of conditioning variables. These plots are arranged in a vector, a matrix or a set of parallel matrix accordingly to the number of conditioning variables. To display adjunct variables different colors and makers are used. This method offers a lot of useful extensions to a standard static graphics. With Trellis Displays one can combine up to eight variables in one plot. The question raises of how far one can obtain equivalent results with the technique of interactive statistical graphics. The method of linking plots together, which is the central point of interactive graphics, seems to be more flexible way of analysing datasets. Trellis Display can be used to present results of interactive statistical graphics in a static way.