The SAS System (originally Statistical Analysis System) is an integrated system of software products provided by
SAS Institute that enables the programmer to perform:
In addition, the SAS System integrates with many SAS business solutions that enable large scale software solutions for areas
such as human resource management, financial management, business intelligence,
customer relationship management and more.
Description of SAS
SAS is driven by SAS programs that define a sequence of operations to be performed
on data stored as tables. Although non-programmer graphical user interfaces to SAS exist (such as the SAS Enterprise Guide), most of the time
these GUIs are just a front-end to automate or facilitate generation of SAS programs. SAS components expose their functionalities
via application programming interfaces, in the form of
statements and procedures.
A SAS program is composed of three major parts, the DATA step, procedure steps (effectively, everything that is not enclosed
in a DATA step), and a macro language. SAS Library Engines and Remote Library Services allow access to data stored in external
data structures and on remote computer platforms.
The DATA step section of a SAS program, like other database-oriented
fourth-generation programming languages such as SQL or Focus,
assumes a default file structure, and automates the process of identifying files to the
operating system, opening the input file, reading the next record, opening the output
file, writing the next record, and closing the files. This allows the user/programmer to concentrate on the details of working
with the data within each record, in effect working almost entirely within an implicit program
loop that runs for each record.
All other tasks are accomplished by procedures that operate on the data set (SAS'
terminology for "table") as a whole. Typical tasks include printing or performing statistical
analysis, and may just require the user/programmer to identify the data set. Procedures are not restricted to only one
behavior and thus allow extensive customization, controlled by mini-languages defined within the procedures. SAS also has an
extensive SQL procedure, allowing SQL programmers to use the system with little additional
knowledge.
There are macro programming extensions, that allow for rationalization of
repetitive sections of the program. Proper imperative and procedural programming constructs can be simulated by use of the "open code" macros or the
SAS/IML component.
Macro code in a SAS program, if any, undergoes preprocessing. At runtime, DATA steps are compiled and procedures are interpreted and run in the sequence they appear in the SAS
program. A SAS program requires the SAS System to run.
Compared to general-purpose programming languages, this
structure allows the user/programmer to be less familiar with the technical details of the data and how it is stored, and
relatively more familiar with the information contained in the data. This blurs the line between user and programmer, appealing
to individuals who fall more into the 'business' or 'research' area and less in the 'information technology' area, since SAS does not enforce (although SAS recommends) a structured,
centralized approach to data and infrastructure management.
The SAS System runs on IBM mainframes, Unix machines,
OpenVMS Alpha, and Microsoft Windows; and code is
almost transparently moved between these environments. Older versions have supported PC-DOS,
the Apple Macintosh, VMS, VM/CMS, Data General AOS and OS/2.
Early history of SAS
SAS was conceived by Anthony J. Barr in 1966.[1] As a North Carolina State
University graduate student from 1962 to 1964, Barr had created an analysis of
variance modeling language inspired by the notation of statistician Maurice
Kendall, followed by a multiple regression program that generated machine code
for performing algebraic transformations of the raw data. Drawing on those programs and his experience with structured data
files[2], he created SAS, placing statistical procedures
into a formatted file framework. From 1966 to 1968, Barr developed the fundamental structure and language of SAS.
In January 1968, Barr and James Goodnight collaborated, integrating new multiple
regression and analysis of variance routines developed by Goodnight into Barr's framework.[3][4]
Goodnight's routines made the handling of basic statistical analysis more robust, and his later implementation (in SAS 76) of the general linear model greatly increased the
analytical power of the system. By 1971, the SAS system was gaining popularity within the academic community. And by 1972,
industry was making use of SAS. One strength of the system was analyzing experiments with missing data, which was useful to the
pharmaceutical and agricultural industries,
among others.
In 1973, John P. Sall joined the project, making extensive programming contributions in
econometrics, time series, and matrix algebra. Other participants in the early years included Caroll G. Perkins, Jolayne W.
Service, and Jane T. Helwig. Perkins made programming contributions. Service and Helwig created the early documentation.[3]
In 1976, SAS Institute, Inc. was incorporated by Barr, Goodnight, Sall, and Helwig.
Components
The SAS system consists of a number of components, which organizations separately license and install as required.
- SAS Add-In for Microsoft Office
- A component of the SAS Enterprise Business Intelligence Server, is designed to provide access to data, analysis, reporting
and analytics for non-technical workers (such as business analysts, power users, domain experts and decision makers) via menus
and toolbars integrated into Office applications.
- Base SAS
- The core of the SAS System is the so-called Base SAS Software, which is used to manage data. SAS procedures
software analyzes and reports the data. The SQL procedure allows SQL programming in lieu of data
step and procedure programming. Library Engines allow transparent access to common data structures such as Oracle, as well as
pass-through of SQL to be executed by such data structures. The Macro facility is a tool for extending and customizing SAS
software programs and reducing overall program verbosity. The DATA step debugger is a programming tool that helps find
logic problems in DATA step programs. The Output Delivery System (ODS) is an extendable system that delivers output in a
variety of formats, such as SAS data sets, listing files, RTF, PDF, XML, or HTML. The SAS
windowing environment is an interactive, graphical user interface used to
run and test SAS programs.
- SAS Enterprise Business Intelligence Server
- Includes both a suite of business intelligence (BI) tools and a platform to
provide uniform access to data. The goal of this product is to compete with Business
Objects and Cognos' offerings.
- Enterprise Computing Offer (ECO)
- Not to be confused with Enterprise Guide or Enterprise Miner, ECO is a product bundle.
- Enterprise Guide
- SAS Enterprise Guide is a Microsoft Windows client application that provides a guided mechanism to use SAS and publish
dynamic results throughout an organization in an uniform way. It is marketed as the default interface to SAS for business
analysts, statisticians, and programmers.
- Enterprise Miner
- A data mining tool.
- ETL
- Provides Extract, transform, load services.
- SAS/ACCESS
- Provides the ability for SAS to transparently share data with non-native datasources.
- SAS/ACCESS for PC Files
- Allows SAS to transparently share data with personal computer applications
including MS Access and Microsoft Office
Excel.
- SAS/AF
- Applications facility, a set of application development tools to create customized applications.
- SAS/ASSIST
- Early point-and-click interface to the SAS system, has since been superseded by SAS
Enterprise Guide.
- SAS/C
- SAS/CONNECT
- Provides ability for SAS sessions on different platforms to communicate with each other.
- SAS/DMI
- A programming interface between interactive SAS and ISPF/PDF applications. Obsolete since version 5.
- SAS/EIS
- A menu-driven system for developing, running, and maintaining an enterprise
information systems.
- SAS/ETS
- Provides Econometrics and Time Series
Analysis
- SAS/FSP
- Allows interaction with data using integrated tools for data entry, computation, query, editing, validation, display, and
retrieval.
- SAS/GIS
- An interactive desktop Geographic Information System for mapping
applications.
- SAS/GRAPH
- Although base SAS includes primitive graphing capabilities, SAS/GRAPH is needed for charting on graphical media.
- SAS/IML
- Matrix-handling SAS script extensions.
- SAS/INSIGHT
- Dynamic tool for data mining. Allows examination of univariate distributions, visualization of multivariate data, and model
fitting using regression, analysis of variance, and the generalized linear model.
- SAS/IntrNet
- Extends SAS’ data retrieval and analysis functionality to the Web with a suite of CGI and Java tools
- SAS/LAB
- Superseded by SAS Enterprise Guide.
- SAS/OR
- Operations Research
- SAS/PH-Clinical
- Defunct product
- SAS/QC
- Quality Control provides quality improvement tools.
- SAS/SHARE
- Is a data server that allows multiple users to gain simultaneous access to SAS files
- SAS/STAT
- Statistical Analysis with a number of procedures, providing statistical information such as analysis of variance, regression,
multivariate analysis, and categorical data analysis.
- SAS/TOOLKIT
- SAS/Warehouse Administrator
- Superseded in SAS 9 by SAS ETL Server.
- SAS Web Report Studio
- Part of the SAS Enterprise Business Intelligence Server, provides access to query and reporting capabilities on the Web.
Aimed at non-technical users.
Terminology
Where many other languages refer to tables, rows, and columns/fields, SAS uses the forms data sets,
observations, and variables respectively. This usage derives from its statistical heritage, and is shared by SPSS, another statistical package.
There are only two kinds of variables in SAS, numeric and character (string). By default all numeric variables are stored as
real. It is possible to reduce precision however. Date and datetime variables are numeric variables that inherit the C tradition
and are stored as either the number of days (for date variables) or seconds (for datetime variables) from an epoch of 1960-01-01
00:00:00.
Features
- Read and write many different file formats.
- Process data in many different formats.
- SAS programming language is a 4th generation programming
language. Actually it is a "3.5 GL" programming language. SAS DATA steps are written in a 3rd-generation procedural
language very similar to PL/I; SAS PROCS, especially PROC SQL, are non-procedural and therefore better fit the definition of a
4GL.
- Many built-in statistical and random number functions.
- Interaction with database products through SQL (and ability to use SQL internally to manipulate
SAS data sets).
- Direct output of reports to CSV, HTML,
PCL, PDF, PostScript, RTF, XML, and more using
ODS.
- Interaction with the operating system (for example, pipelining on Unix and Windows and DDE on Windows).
- Fast development time, particularly from the many built-in procedures.
- Hundreds of built-in functions for manipulating character and numeric variables.
- An integrated development environment.
- Dynamic data-driven code generation using the SAS Macro language.
- Can process files containing millions of rows and thousands of columns of data.
- University research centers often offer SAS code for advanced statistical techniques, especially in fields such as Political
Science, Economics and Business Administration.
Example SAS code
SAS uses data steps and procedures to analyze and manipulate data. By default, a data step iterates through each observation
in a data set (sort of like every row in a SQL table).
This data step creates a new data set BBB that includes those observations from data set AAA that had charges greater than
100.
data BBB;
set AAA;
if charge > 100;
run;
Procedures that can summarize data are available in SAS. The proc freq procedure shows a frequency distribution of a
given variable in a data set.
proc freq data=BBB;
table charge;
run;
SAS features a macro language, which can be used to generate SAS code. For instance, the above example could be re-used in
many pieces of code by rewriting it as a macro:
%macro freqtable(table, variable);
proc freq data = &table;
table &variable;
run;
%mend freqtable;
%freqtable(BBB, charge)
SAS also features SQL, which can be used to query SAS datasets or external database tables accessed with a SAS libname engine.
For example, duplicate records could be extracted from a table for analysis:
proc sql;
create table dup_recs
from your_dataset d,
(select count(*), id
from your_dataset
group by id
having count(*) > 1
) t1
where d.id=t1.id;
quit;
SAS has a useful feature where it can display the queried information. The proc print procedure is used for this:
proc print data=BBB;
run;
Version history
SAS 71
SAS 71 was the first limited release of the system. The first manual for SAS was printed at this time, approximately 60 pages
long[5]. The DATA step was implemented. Regression and
analysis of variance were the main uses of the program.
SAS 72
This more robust release was the first to achieve wide distribution. It included a substantial user's guide, 260 pages in
length[6]. The MERGE statement was introduced in this
release, adding the ability to perform a database JOIN on two data sets[7]. This release also introduced the comprehensive handling of missing data[8].
SAS 76
SAS 76 was a complete system level rewrite, featuring an open architecture for adding and extending procedures, and for
extending the compiler[9]. The INPUT and INFILE statements
were significantly enhanced to read virtually all data formats in use on the IBM mainframe[10]. Report generation was added through the PUT and FILE statements[11]. The capacity to analyze general linear models was added[12].
79.3 - 82.4
1980 saw the addition of SAS/GRAPH, a graphing component; and SAS/ETS for econometric and time series analysis. In 1981
SAS/FSP followed, providing full-screen interactive data entry, editing, browsing, retrieval, and letter writing.
In 1983 full-screen spreadsheet capabilities were introduced (PROC FSCALC).
For IBM mainframes, SAS 82 no longer required SAS databases be DSORG=DAU, because SAS 82 removed location-dependent
information from databases. While this may seem trivial, it eliminated a major headache in administering SAS--that restoring a
SAS disk-base database from tape no longer required restoring the entire volume, then copying the database to another
location.
Version 4 series
In the early 1980s, SAS Institute released Version 4, the first version for non-IBM computers. It was written mostly in a
subset of the PL/I language, to run on several minicomputer
manufacturers' operating systems and hardware: Data General's AOS/VS, Digital Equipment's VAX/VMS, and Prime Computer's
PRIMOS. The version was colloquially called "Portable SAS" because most of the code was portable, i.e., the same code would run
under different operating systems.
Version 5 series
Version 6 series
Version 6 represented a major milestone for SAS. While it was superficially similar to the user, the major change was "under
the hood", where the software was rewritten. From its FORTRAN origins, followed by PL/I and
mainframe assembly language; in version 6 the SAS System was rewritten in
C, to provide enhanced portability between operating systems, as well as access to an increasing pool of C
programmers compared to the shrinking pool of PL/I programmers.
This was the first version to run on UNIX, MS-DOS and Windows platforms. The DOS versions were incomplete implementations of
the Version 6 spec: some functions and formats were unavailable, as were SQL and related items such as indexing and WHERE
subsetting. DOS memory limitations restricted the size of some user-defined items.
The mainframe version of SAS 6 changed the physical format of SAS databases from "direct files" (DSORG=DA) to "flat files"
(DSORG=PS,RECFM=FS). The practical benefit of this change is that a SAS 6 database can be copied from any media with any copying
tool.
In 1984 a project management component was added (SAS/OR?).
In 1985 SAS/AF software, econometrics and time series analysis (SAS/DMI) component, and interactive matrix programming
(SAS/IML) software was introduced. MS-DOS SAS (version 6.02) was introduced, along with a link to mainframe SAS.
In 1986 Statistical quality improvement component is added (SAS/QC software); SAS/IML and SAS/STAT software is released for
personal computers.
1987 saw concurrent update access provided for SAS data sets with SAS/SHARE software. Database
interfaces are introduced for DB2 and SQL-DS.
In 1988 MultiVendor Architecture (MVA) concept is introduced; SAS/ACCESS software is released. Support for UNIX-based hardware
announced. SAS/ASSIST software for building user-friendly front-end menus is introduced. New SAS/CPE software establishes SAS as
innovator in computer performance evaluation. Version 6.03 for MS-DOS is released.
6.06 for MVS, CMS, and OpenVMS is announced in 1990. The same year, the last MS-DOS version (6.04) is released.
Data visualization capabilities added in 1991 with SAS/INSIGHT software.
In 1992 SAS/CALC, SAS/TOOLKIT, SAS/PH-Clinical, and SAS/LAB software is released.
In 1993 software for building customized executive information systems (EIS) is introduced. Release 6.08 for MVS, CMS, VMS,
VSE, OS/2, and Windows is announced.
1994 saw the addition of ODBC support, plus
SAS/SPECTRAVIEW and SAS/SHARE*NET components.
6.09 saw the addition of a data step debugger.
6.09E for MVS.
6.10 in 1995 was a Microsoft Windows release and the first release for the Apple Macintosh. Version 6 was the first, and last
series to run on the Macintosh. JMP, also produced by the SAS Institute, is
the software package the company produces for the Macintosh.
Also in 1995, 6.11 (codenamed Orlando) was released for Windows 95, Windows NT, and UNIX.
6.12 were Unix and Microsoft Windows releases (and more?)
(Some of the following milestones in this sub-section may belong under version 7 or 8.)
In 1996 SAS announces Web enablement of SAS software. Scalable performance data server is introduced.
In 1997 SAS/Warehouse Administrator and SAS/IntrNet software goes into production.
1998 sees SAS introduce a customer relationship management (CRM) solution, and an ERP access
interface — SAS/ACCESS interface for SAP R/3. SAS is also the first to release OLE-DB for OLAP and releases HOLAP solution.
Balanced scorecard, SAS/Enterprise Reporter, and HR Vision are released. First release of SAS Enterprise Miner.
1999 sees the releases of HR Vision software, the first end-to-end decision-support system for
human resources reporting and analysis; and Risk Dimensions software, an end-to-end risk-management solution. MS-DOS versions are
abandoned because of Y2K issues and lack of continued demand.
In 2000 SAS shipped Enterprise Guide and ported its software to Linux.
Version 7 series
The Output Delivery System debuted in version 7; as did long variable names (from 8 to 32 characters); storage of long
character strings in variables (from 200 to 32,767); and a much improved built-in text editor, the Enhanced Editor.
Version 7 saw the synchronisation of features between the various platforms for a particular version number (which previously
hadn't been the case).
Version 7 was a precursor to version 8. It was believed SAS Institute released a snapshot from their development on version 8
to meet a deadline promise. SAS Institute recommended that sites wait until version 8 before deploying the new software.
Version 8 series
Released about 1999; 8.0, 8.1, 8.2 were Unix, Microsoft Windows, CMS (z/VM) and z/OS releases. Key features: long variable names, Output Delivery System
(ODS).
Version 9 series
In version 9, SAS Institute added the SAS Management Console, parallel processing,
JavaObj, ODS OO (experimental as opposed to alpha), and National Language Support.
Again the SAS Institute recommended sites delay deployment until 9.1.
SAS Version 9 is running on Windows (32 & 64 bit), Unix (64 bit), Linux, and z/OS. Support for CMS (z/VM) was dropped.
SAS 9.1 was released in 2003.
SAS 9.1.2 was released in 2004.
SAS 9.1.3 was released in 2005.
SAS 9.1.3 Service Pack 4 is the latest release (April 2006). The latest hotfix level for this service pack is E9BX02
(September 2007)
SAS 9.2 is the next release[1] and was demonstrated at the SUGI31 Conference in March 2006[2].
There are several important additions to base SAS in Version 9. The new hash object now allows functionality similar to the
MERGE statement without sorting data or building formats. The function library was enlarged, and many functions have new
parameters. Perl Regular Expressions are now supported, as opposed to the old "Regular Expression" facility, which was
incompatible with most other implementations of Regular Expressions. Long format names are now supported.
Criticism
SAS has been criticized for its relatively poor graphics when compared with other statistical software packages. With the
release of an experimental extension to SAS 9.1, the graphics have improved significantly. The enhanced graphics are not provided
by default, though, and usually get stored in many pieces.
SAS has also been criticized for its costs, especially when compared to its open source competitors such as R
SAS is considered to be several years behind competitor products when it comes to the implementation of statistical
algorithms
SAS has been criticised for its syntax unlike any other (popular) programming languages.
Another criticism of SAS is the excessive amount of whitespace in its output.
External links
Notes
- ^ Greenberg & Cox, et al. 1978:181. Reference to the creation of SAS by
Barr in 1966.
- ^ Barr contributed to the development of the NIPS Formatted File System while working for IBM at the Pentagon from 1964 - 1966. FFS was one of
the first data management systems to take advantage of files with a defined structure for efficiencies in data storage and
retrieval.
- ^ a b (Barr & Goodnight, et al. 1976:"The SAS Staff") Attribution of
contributions to SAS 72 and SAS 76 to Barr, Goodnight, Service, Perkins, and Helwig.
- ^ (Barr & Goodnight et al. 1979:front matter) Attribution of the
development of various parts of the system to Barr, Goodnight, and Sall.
- ^ (Barr & Goodnight 1971)
- ^ (Service 1972)
- ^ (Service 1972:47-49)
- ^ (Service 1972:28,65,67,etc.)
- ^ (Barr & Goodnight, et al. 1979) This programmer's guide facilitated the
extension of the SAS system through its open interface.
- ^ (Barr & Goodnight, et al. 1976:11-15)
- ^ (Barr & Goodnight, et al. 1976:38-44)
- ^ (Barr & Goodnight, et al. 1976:127-144)
References
- SAS Company History
- Barr, Anthony J., Goodnight, James H. SAS, Statistical Analysis System, Student Supply Store, North Carolina State
University, 1971. OCLC
5728643
- Barr, Anthony J., Goodnight, James H., Sall, John P., Helwig, Jane T. A User's Guide to SAS 76, SAS Institute, Inc.,
1976. ISBN 0-917382-01-3
- Barr, Anthony J., Goodnight, James H., Sall, John P., Helwig, Jane T. SAS Programmer's Guide, 1979 Edition, SAS
Institute, Inc., 1979. OCLC
4984363
- Cody, Ron and Ray Pass. SAS Programming by Example. 1995. SAS Institute.
- Delwiche, Lora D. and Susan J. Slaughter. The Little SAS Book. 2003. SAS Institute.
- McDaniel, Stephen and Hemedinger, Chris. SAS for Dummies. 2007. Wiley.
- Greenberg, Bernard G.; Gertrude M.
Cox & David D. Mason et al. (1978), Nourse, E. Shepley, ed., "Statistical Training and Research: The University of North Carolina System", International Statistical Review
46: 171-207, <http://links.jstor.org/sici?sici=0306-7734(197808)46%3A2%3C171%3ASTARTU%3E2.0.CO%3B2-S>
- Service, Jolayne A User's Guide to the Statistical Analysis System., Student Supply Stores, North Carolina State
University, 1972. OCLC
1325510
This entry is from Wikipedia, the leading user-contributed encyclopedia. It may not have been reviewed by professional editors (see full disclaimer)
