RSS Matters

By Dr.Karl Ho, Research and Statistical Support Services Manager

New Generations of Statistical Applications

As the clock clicks closer to the next millennium, we are reminded that we have to face a new generation of computers and software applications. Yes, they are here! Beginning this month, we introduce the new wave of "next generation" statistical packages. We start in this issue with SAS 7.0, which is among the biggest innovators in my opinion. But let's first examine what we have received lately: 

Most Current Versions of Statistical Applications at RSS

red font - newly arrived
* under evaluation
# to be installed

SAS 7.0

Writing an evaluation of the SAS products is a formidable task. It is a love-hate, hate-love job. Delights and excitements abound, amid the discovery of features that only exist at the farthest borders of my imagination. It is, however, equally painfully to recognize how little I know about this system and how ignorant I am in front of this fast-expanding information world.

Disclaimer: This will not be a full scale evaluation of SAS 7. Reason 1: the limited room in this column. Reason 2: SAS Institute is going to release the next version, SAS 8, in the second half of 1999. Reason 3: can anyone out there do it?

All that said, it is still a pleasure to explore this latest version of the SAS system. SAS 7 is the first production release of a project code-named Nashville. The project is designed to "provide an umbrella of technology to simplify the processes of data access, data management, and data exploitation."¹ The latest version of SAS should not be considered a Statistical pack, nor a number crunching application. It is better recognized as a data system because of its capabilities of data management, analysis, storage and delivery.  If you knew SAS before, be ready to change your presumptions about SAS being too difficult to use, too data-unfriendly, or too dull in its interface. SAS takes pains to change its outlook and make it as user-friendly as most other Windows applications yet retain and, in some areas, enhance the advantages of the programmability of the SAS language. Advanced programmers and beginners alike can enjoy this new version.  

New Outlook

On top of the three default windows (Program Editor, Log, Output), two more windows are added: the Explorer and the Results window. The former, which was hidden within SAS/Desktop in version 6, allows access to all SAS data objects including data libraries, data members and catalogs plus file shortcuts. The Results window demonstrates the new Web-ready feature that automatically translates all SAS output into HTML, including charts. Click here to see an example of a sample output in HTML format. This example demonstrates the output of a correlation analysis that exposes the relationship between baseball players' salaries and their homerun and hits records. The charts are scatter plots of each bivariate pair with 0.95 confidence ellipse.

Behind the scene of this translation is a new SAS feature called Output Delivery System (ODS).  Not only does ODS transform output into HTML and Web-ready graphics, this new system can turn output into SAS data sets. The ODS codes that perform the analysis are as follows:

ods listing;
ods html
path="C:\My Documents\My SAS Files\analyst_projects\HTML4950" (url=none)
frame="FRAME.html"
body="BODY.html"
contents="CONTENTS.html"
dev=gif nogtitle nogfootnote;

Alternately, the process can be performed via a SAS module called Analyst, which allows users to perform data analysis via a point-and-click approach. Details of Analyst will be covered later in this article.

Data Access

Data access to other data formats becomes much easier with SAS/Access. The Import Wizard (File--Import) leads all the way to read data into SAS. New compatible formats include Excel 97 and Access 97 tables.  (See Figure 1).

Figure 1. Import Wizard in SAS 7

New Menu

Another visible change in SAS 7 is the rearrangements on the menu. All analyses and related functionalities are organized under Solutions now. Table 1 and Figure 2 illustrate the new paradigm:

Table 1. Functionalities Listed Under Solutions 

Figure 2. New arrangements of modules under the SAS 7 menu

Particularly noteworthy is the Analyst Module, which used to be an add-on in Version 6. It has become a member under the Solutions category. The enhanced version of Analyst has a new look and organizational structure with better access to data, output and graphics. It allows for production of charts such as histograms, pie charts and bar graphs with just a couple of clicks. In addition, Analyst is empowered with new analytical modules including survival analysis, repeat measurements, mixed model and power estimations. Other existing functionalities include ANOVA, regression and principal component analysis and canonical correlations.

Other Statistical procedures added to SAS 7 include:

• Partial Least Square
• Spatial prediction
• GEE analysis in GENMOD
• PROC TLOGISTIC

Other new features

Support for long names has been built in so names of variables, libraries, and catalog entries can be up to 32 mixed case characters and labels can be up to 256 characters. Also, users will be able to get access to the full version of SAS documentation via an Internet browser.

Although it is far from being complete, we hope this review provides you with glimpses into the new looks of SAS 7 and some new features that characterize the new generation of the Statistical applications. In the next issue of Benchmarks Online, we will provide more details on SAS 7 and SPSS 9.0

Figure 3. A Time Series Plot in SAS 7

Figure 4.  A Surface Plot in SAS 7

¹Version 7 of the SAS® System: The Initial Release of the Nashville Project, a publication from SAS introducing the project (http://www.sas.com/service/techsup/version7.pdf)