Graphics capabilities and applications of the SAS statistical software system

SAS System

SAS contains an excellent graphics package for producing graphics and visuals to enhance the display and reporting of statistical information. The focus of this study is to examine the graphics uses of the SAS system and to provide a description for each use of graphics with the SAS system and then to compare the SAS system with other graphics packages available in today's market. Toward this end, this work will conduct a review of literature in this area of study.

(1) What are the various types of graphics that can be performed using the SAS system?

(2) What are the applications of SAS graphics?

(3) How does SAS graphics differ from other graphics packages?

The work of Phil Spector reports that the SAS system was developed in the early 1970s decade at North Caroline State University and was intended originally for use by management and analysis of agricultural field experiments. (paraphrased) SAS is reported as the most widely used statistical software. SAS stands for 'Statistical Analysis System'. SAS products are reported by Spector ( ) to include those stated as follows:

(1) Base SAS -- data management and basic procedures;

(2) SAS/STAT -- statistical analysis;

(3) SAS/GRAPH -- presentation quality graphics;

(4) SAS/OR -- operations research;

(5) SAS/ETS -- Econometrics and Time Series Analysis;

(6) SAS/IML -- Interaction matrix language;

(7) SAS/AF -- applications facility (menus and interfaces); and (8) SAS/QC -- quality control. (Spector, nd, p.2)

Spector reports that there are other "specialized products for spreadsheets, access to databases, connectivity between different machines running SAS" as well as others. (nd, p.2) Most SAS programs have two primary components: (1) the data step(s); and (2) the procedure step(s). (Spector, nd, p.3) The data step "reads data from external sources, manipulates, and combines it with other data set and print reports. The data step is used to prepare the data for use by one of the procedures." (Spector, nd, p.6)

SAS is reported as being "very lenient about the format of its input -- statements can be broken up across lines, multiple statements can appear on a single line and blank spaces and lines can be added to make the program more readable." (Spector, nd, p.6) The procedure steps are such that "perform analysis on the data, and produce outputs" often in large amounts. (Spector, nd, p.6) The most effective learning strategy for SAS is concentration on the "details of the data step and learn the details of each procedure" as there is a need for them. (Spector, nd, p.6)

It is reported by UCLA Academic Technology Services that SAS graphics include:

(1) Graph-N-Go

(2) SAS/INSIGHT;

(3) SAS/ANALYST; and (4) SAS/PROCS. (UCLA Academic Technology Services, 2012, p.1)

I. Graph-N-Go

Graph-N-Go is reported as being primarily for reporting and to have the strong point of "flexibility to save a plot in various formats, including graphic format and html format. Its weak point is that it only supports a few graph types." (UCLA Academic Technology Services, 2012, p.1) The Graph-N-Go reporting creates a graphic that appears in the following illustration labeled Figure 1.

Figure 1 - Graph-N-Go

Source: UCLA Academic Technology Services (2012)

II. SAS/Insight

SAS/Insight is reported as a package that can be used in the exploration of variable and relationships among variables. The strong point is that this package offers a great amount of detailed information concerning variables including "such as univariate statistics." (UCLA Academic Technology Services, 2012, p.1) In addition, all of the graph files can be saved to .gif files or other graphic files. The limitation is that it is difficult to modify the color of style of the graphs. However, its strong feature is its interaction feature, which enables exploration of data "both graphically and analytically." (UCLA Academic Technology Services, 2012, p.1) The following illustration shows the SAS/Insight graphics example:

Figure 2 SAS/Insight Graphic (Rotating Plot, YZX)

Source: UCLA Academic Technology Services (2012)

III. SAS/Analyst

SAS Analyst attempt to explore and report and is described as an excellent place to begin since many types of graphs can be creating and the SAS code for creating the plots is created as well and the settings can be changed and code modified if it is needed. This is true for graphics and statistical analysis. The following illustration shows an SAS/Analyst graph.

Figure 3 SAS/Analyst Graph

Source: UCLA Academic Technology Services (2012)

The options used to generate this analysis are reported to be that stated as follows:

title;

footnote;

goptions ftext=SWISS ctext=BLACK htext=1.0 cells;

goptions colors=(red green blue cyan purple tan pink orange brown yellow plum peru salmon lime);

axis1 label=(a=90 r=0);

pattern value=solid;

*** Produce bar charts ***;

proc gchart data=Work.Hsb2;

vbar3d WRITE

/ description="Vertical Bar Chart of WRITE"

frame woutline=1

type=FREQ

group=female subgroup=GROUP

coutline=CX008080

cframe=CXF7E1C2

inside=FREQ

run;

goptions ftext= ctext= htext=;

quit;

V. Comparison to Other Graphics Packages

There are other graphics packages including SigmaPlot, PlotIt, Axum and others "where graphs are produced from spreadsheet like data. Annotations are placed on the graph in an interactive fashion. The graph type is chosen through a graphical user interface, and so forth. In SAS/GRAPH graphics are produced usually by programming statements in the PROGRAM EDITOR and executing them (there are other ways to create graphs in SAS, e.g. SAS/INSIGHT and SAS/SPECTRAVIEW). The definition of axes, annotations, notes on the graph, the data to be used, the size and type of symbols and lines are all part of the program." (UCLA Academic Technology Services, 2012, p.1) Novice users find this challenging however, it is a powerful feature enabling the user to gain "complete control over the appearance of a graph, it results in a storable program that can be executed at a later time for a different data set, producing essentially the same graph with new data. One can create multiple graphs from a data set with annotations that are placed identically, can change their values depending on the contents of a data set without user intervention. Rather than storing a huge graphics file" the individual can choose to save a small SAS program that creates the desired graphic upon execution. (UCLA Academic Technology Services, 2012, p.1)

Goptions are reported to be the graphics options that give definition to the output device in terms of the formatting of the graphics and whether there is a screen preview display, or the data is written to a file, or in regards to the graph size or orientation, the setting of options enables the individual to trim down the procedural statements, code in axis and legend definitions. (UCLA Academic Technology Services, 2012, paraphrased)

The work of Friendly (nd ) reports that statistical graphics has as its goal gaining insight into the data and states that SAS/GRAPH procedures alone "often do not go far enough in providing the tools for the most effective visual displays and reports on some useful graphic displays that cannot be produced by the basic SAS/GRAPH procedures and shows how they can be constructed by general macro programs applicable to any dataset." (p.1) Friendly (nd) states that the basic SAS/GRAPH procedures make provision of "a wide range of methods for graphic display of data -- including charts (pie charts, bar charts, 3-D block charts) scatterplots (with points, needles, bubbles, regression lines, etc.), 3D plots (scatter, surface or contour) as well as facilities to display geographical data on a variety of maps." (Friendly, nd, p.2) Friendly writes that making the provision for the basic construction of custom graphic displays are "Annotate, PROC GREPLAY and SAS/IML." (nd, p.2) Each of these are explained as follows:

(1) SAS/IML -- a full-featured language for statistical computation and graphics on its own. Especially useful in applications that require computation on a data seta as a whole, such as finding row and column means or which are naturally expressed in matrix operations, such as calculating a data elliprse for a bivariate scatterplot." (Friendly, nd, p.2) SAS/IML is reported to contain a "powerful and relatively complete set of high-level graphics functions that can be used in the implementation of "customized graphic displays." (Friendly, nd, p.2)

(2) PROC GREPLAY -- the tool of choice for graphs comprised of separate panels including scatterplot matrices, plots of means for multi-way designs or displays combining plots, charts, and explanatory text." (Friendly, nd, p.2)

(3) DATA Step Graphics -- a new addition to SAS/GRAPH, the DATA Step Graphics Interface (DSGI) enables the creation of graphics output directly within a DATA sep or from within an SCL application. (Friendly, nd, p.2)

(4) SAS Macro Language -- organization of the SAS System into DATA steps and PROC steps is "convenient for data analysis. The SAS Macro Language enables the packaging of any number of SAS program steps into only one single unit. This makes provision of a method for extending and customizing graphic displays, resulting in the creation of general procedures that can be applied to any set of data. (Friendly, nd, p.2)

(5) Custom Tools -- SAS macros can reportedly be used in creation of "general purpose, modular tools for graphic programming. One begins with a simple version for a repetitive task and adds features as needed. The macros are stored in an auto call library or a compiled macro library and kept available for all applications. (Friendly, nd, p.2)

Friendly states in a separate work in writing entitled "Multidimensional Arrays in SAS/IML Software" that SAS/IML "contains a rich variety of matrix operations and built-in functions for statistical computation and graphics. Other array-oriented programming languages, such as Mathematics, APL and MATLAB, provide some of these facilities, but the combination of high-level operations, statistical functions and graphics in SAS/IML is hard to beat for most of the graphics applications that are under development. A comparison of SAS and SPSS products with R. Packages nd Functions reports that R. is "open source software for statistics and graphics which is available for free. R is reported to not only do almost everything done by SAS and SPSS but as well to do "quite a bit more ." (Friendly, nd, p.2) The comparison of the SAS, SPSS and R. Package is listed in the table, which follows.

Table 1 -- Comparison of SAS, SPSS and R. Package

Source: Muenchen (nd)

Patel and Caira (nd) report that the HTML device driver can be used to create a simple scrollable graph that can be viewed on any viewer. The HTML device driver is reported to create one or more gif files and the HTML file that refers to the GIF images created. It is really quite understandable as explained by Patel and Caira (nd) who state that in order to create a series of graphs the following is used:

GOPTIONS DEVICE=HTML

This creates a default HTML file called index.html that can be viewed on any browser. Additionally created is a GIF file for every graph that is produced by the procedure." (nd) Next one can utilize a webframe driver, which is reported to work just as any HTML device but with a layout that is more advanced, which then creates GIF files and several HTML files and the GIF images are displayed with thumbnail links breaking the layout by default into two views. Those with extensive knowledge of HTML are able to use output from SAS/GRAPH software with drill-down options in working with data combined through syntax with SAS DATA steps and micros to generate the HTML output. IMAGEMAP is used in generating GCHART, GPLOT, or GMAP output.

Analysis

It is clear from the literature reviewed in this study that SAS graphics package is capable of performing much more graphical representation than any product on today's market. Not only has this study demonstrated that SAS graphics system is able to produce more graphical representations that are more effective but as well, that SAS system graphics can be modified and amended after input where other graphics packages cannot be amended after input of data. Multivariate and univariate statistical graphics are also available in SAS system graphics package as well. The real value however of the SAS graphics package are the cuts in time and display benefits available to those who care to learn this particular system graphics for statistical modeling.

Conclusion

There are many products available however, none that are able to match the overall potential such as found available using SAS. SAS may be a little seemingly tedious to learn but is reported by experts to be the preferred method for visualizing graphical statistical modeling data.