The Graphical Interface
From StatSoft Of all the Windows-based graphical packages surveyed, Statistica was the most impressive. It has an almost dizzying array of graphics options. It undoubtedly has the quickest access to the most powerful graphics of any package. There were, of course, some weaknesses and difficulties, which are discussed below. However, it is very obvious that this is a prototype for where graphics packages of the information age will be heading. UNIQUE GRAPHICS FEATURES: Statistica has a graphical feature for plotting the joint distribution of two continuous variables in a bivariate histogram. It also has interactive data links that allow the user to highlight outliers for removal or to explore multidimensional relationships with other variables of interest. UniStat was the only other package that had these two features. Statistica also has an expanded type of polar plot, which provides a novel means of presenting standard two-dimensional box-plot information in a circular format. This is very pleasing to the eye compared to standard box plots. Its spectral planes option in the three-dimensional graphics displays provides a novel method of compressing information in a three-dimensional plot. Statistica also had the best graphical accompaniment to multivariates analyses of any package, including very good multivariate diagnostics of normality, outliers and residuals. STRENGTHS: Statistica has many strengths, including very good iconic graphics for representing cases or units of observation as multidimensional symbols. Statistica also has entensive ternary plot options and matrix plot options (also called sploms). Moreover, there are very good three-dimensional graphics options that include spinning options and animated stratification. Animated stratification allows the user to view a three-dimensional graph dynamically (surface or box plot) in a layer-by-layer sequential fashion, revealing the inner portions of a complicated three-dimensional plot. It also has very good graphics to go with experimental, factorial designs. WEAKNESSES: The labyrinth of screens for producing graphs can be hard to follow intuitively. There is definitely a learning curve associated with this package. Also, the reference manual is filled with cross referencing, which sometimes makes finding the actual procedures or commands difficult. On large datasets, I found that it could not handle very fine scale, three-dimensional graphics, but then again, no other Windows-based package tested could either. Even with what seems to be a never-ending abundance of options, there still appeared to be holes or dead ends in some of the graphical procedures. LIKELY USERS: This package would be good for anyone who is really dedicated to graphical representations of data and who performs a wide variety of analyses. However, for those individuals who get vertigo with multiple Windows formats, this could be a very stressful package with which to work. Even so, this was the most well-rounded and powerful Windows package on the market.
|
|||