Interactive color graphics for multivariate data

Meaningful display of data structure in more than two dimensions is a challenge to statistical graphics. The Analysis of Large Data Sets Project (ALDS) utilizes a VAX 11/780 and a RAMTEK 9400 high resolution color graphics display device to evaluate potential tools such as motion, color, glyphs and stereo 3-D for display of pseudo higher-dimensional data. This paper describes some of the ALDS research on the use of color and glyphs. Color is helpful but not essential. It is particularly useful for accent, for redundant expression of geometric cues and for identifying categories. Carefully selected color sets can display an ordered scale of a quantitative back variable. Glyphs can display quantitative back variables in a higher-dimensional scatterplot or summary statistics on back variables in each cell of a plot cellulated on two front variables. Examples illustrate the simultaneous use of color and glyphs in these two situations. The ALDS Project has developed a general purpose glyph description language (GDL) which is embedded in an expanded MINITAB statistics package. GDL provides easy definition of arbitrary glyphs and flexible binding of glyph characteristics to back variables. Thus, GDL is a powerful tool for the evaluation of glyphs in an interactive graphics environment.

[1]  Stephen E. Fienberg,et al.  Graphical Methods in Statistics , 1979 .

[2]  G Kolata,et al.  Computer graphics comes to statistics. , 1982, Science.

[3]  E. Anderson A SEMIGRAPHICAL METHOD FOR THE ANALYSIS OF COMPLEX PROBLEMS. , 1957, Proceedings of the National Academy of Sciences of the United States of America.

[4]  Ramanathan Gnanadesikan,et al.  Methods for statistical data analysis of multivariate observations , 1977, A Wiley publication in applied statistics.