SAS Field Guide
When you work with SAS, you write a SAS program or "SAS code". Your
SAS code is a series of instructions, written in the "SAS language".
These instructions typically start-out with some brief specifications
on output constraints (for example, page size, line size, whether you
want the output centered or left-justified, etc). These specifications
are usually followed by some definitions detailing where to find data
files, what format the data files are written in, what to call the
files, and how to input the data. After this, normally one would write
instructions to read the data into a SAS dataset (a form that SAS can
use while it's running). As the data are read into a SAS dataset,
variables optionally may be recoded (modified) or other variables can
be calculated and added into the dataset. Any of a number of "data
manipulations", including very complicated restructuring can be done as
data are input into a SAS dataset. After the instructions to create a
SAS dataset come instructions to tell SAS which statistical procedures
to perform and with what specifications.
The two steps of creating a SAS dataset and running statistical
procedures are divided in SAS as DATA steps and PROCedures. Data steps
begin with a DATA statement and end with a run; statement or, if
there is no run; statement, then it ends at the next DATA step
or PROC. These DATA and PROCedure steps can be interspersed throughout
a SAS program to accomplish several different analyses on any of a
number of different data files or SAS datasets. But it is very
important to realize that you cannot combine a PROCedure within a
DATA step, and you cannot perform data manipulations within a
PROCedure. Higher level functions (eg. output) of some SAS
PROCedures allow one to create new datasets from the results of the
PROCedure, but aside from such special options, DATA and PROC steps
must be written as distinct operations in your SAS code.
Most programming in SAS is simple and nearly understandable as
"structured English". SAS will handle most of the details of looping
through the observations of the data, so you don't need to write "do
loops" unless you intend to do some data restructuring or iterative
recoding. The statistical analyses that SAS can perform work somewhat
like "function calls" or "procedures" in other programming
languages--the complicated mathematics of an analysis are already
written for you as part of the PROCedure--you merely need to "call" the
PROCedure and specify which dataset to use and with what parameters.
When you're done writing your SAS code, you submit
the code, or set of instructions, to SAS. SAS will read and interpret what
you wrote and perform the tasks you specified.
Although it is possible to include your data as part of your SAS code, the
usual method is to keep your data in a separate file and write instructions
in your SAS code for reading the data from an external data file. A SAS
session may be composed of several components (see diagram below). You
write your SAS code as one file and your data reside in a file of their
own. You submit your SAS code for processing to "The SAS System", a series
of programs, interpreters, and libraries that make up "SAS". The SAS system
interprets your code and carries-out the tasks you specified, including
reading-in any data files you may have specified in your SAS code. As
SAS finishes processing your code, the results are written to an OUTPUT
window (or directly to a listing [*.lst] file, if running batch) and a
transcript of the SAS session, including any error messages, warnings, and
notes are written to a LOG window (or *.log file, if running batch).
+---------------------------+ +---------------------------+
| | | |
| SAS Program or "Code" | | Data file on disk or |
| that you write | | tape or CD-ROM, etc |
| | | |
| | | |
+---------------------------+ +---------------------------+
| |
V V
+---------------------------+ +-------------------+
| | | |
| | | LOG file |
| The SAS System |--->| Includes errors, |
| | | warnings, notes. |
| | | |
+---------------------------+ +-------------------+
|
V
+---------------------------+
| |
| |
| Output (Results) |
| |
| |
+---------------------------+
Normally, you will work with three windows: The Program Editor, The
LOG, and The OUTPUT windows. In the Program Editor window you create, open,
or modify a SAS program. LOG window contains the execution log of a
session, and the output of SAS programs are in the OUTPUT window.
- Program Editor: Your SAS code goes here.
- LOG: A log of your SAS session. This is where notes, warnings,
and errors are reported.
- OUTPUT: This is where your results appear.
When you are ready to execute a SAS program that you prepared in the
Program Editor window, SAVE it to your floppy disk first, then from
the LOCAL pull-down menu choose SUBMIT. Alternatively, you can click on
the submit button, which looks like a little stick figure running. As the
program is executed, a transcript of the session--with any error messages--will
be written to the LOG window. If the program runs correctly,
the OUTPUT window will display the results. Always review the
LOG window after running your program--even if it appears that your
program ran correctly. Some serious errors, such as misread or
incorrectly aligned data, may go unnoticed otherwise. Always check
for error messages and warnings in the LOG window.
The Program Editor window will become empty after you submit a job. If
you forgot to save it, you can choose RECALL TEXT from the LOCAL
pull-down menu (or press F4).
When you save your work, you should follow these conventions:
- The name of a SAS source code should have extension .sas;
- The log of a session should have extension .log;
- The output should have extension .lst.
SAS is portable, which means your SAS applications function the same,
look the same, and produce the same results on any hardware platform
and under any operating system. You can develop SAS applications in
one environment and run them in other environments without rewriting
any code except for a few lines where you may be specifying external
data files.
Sample SAS Program
Here's a sample SAS program, modified slightly from a suggestion by
Haihong Li of the Penn State Department of Statistics. The parts in
blue are actual SAS commands, or program steps. The parts written in
black are "comments", or notes, to ourselves (ie. humans) about the
program. SAS ignores--skips over--any comments within a program and
works only on actual program steps:
OPTIONS PS=66 LS=72;
/***********************************************************************
* The above OPTIONS statement is a SAS system option, it sets the
* page size to be 66 lines, and 72 characters per line.
***********************************************************************/
/***********************************************************************
* Filename: fitness.sas
* Written by: Haihong Li
* Date: June 10, 1996
* A small program to demonstrate the basic structure of a SAS program
* that contains data as part of the program itself.
***********************************************************************/
/* Create a dataset called "fitness": */
DATA fitness;
INPUT name $ weight waist pulse chins situps jumps;
CARDS;
Hodges 191 36 50 5 162 60
Kerr 189 37 52 2 110 60
Putnam 193 38 58 12 101 101
Roberts 162 35 62 12 105 37
Blake 189 35 46 13 155 58
;
RUN;
/* Sort the dataset "fitness" by name: */
PROC SORT;
BY name;
RUN;
/* Print out the sorted dataset: */
PROC PRINT data=fitness;
title 'Fitness data';
RUN;
/* Get some descriptive statistics: */
PROC MEANS maxdec=1 data=fitness;
RUN;
If you run it, you'll get the output like this:
Fitness data 21:24 Monday, June 10, 1996 1
OBS NAME WEIGHT WAIST PULSE CHINS SITUPS JUMPS
1 BLAKE 189 35 46 13 155 58
2 HODGES 191 36 50 5 162 60
3 KERR 189 37 52 2 110 60
4 PUTNAM 193 38 58 12 101 101
5 ROBERTS 162 35 62 12 105 37
Fitness data 21:24 Monday, June 10, 1996 2
Variable N Mean Std Dev Minimum Maximum
-------------------------------------------------------------------
WEIGHT 5 184.8 12.9 162.0 193.0
WAIST 5 36.2 1.3 35.0 38.0
PULSE 5 53.6 6.4 46.0 62.0
CHINS 5 8.8 5.0 2.0 13.0
SITUPS 5 126.6 29.4 101.0 162.0
JUMPS 5 63.2 23.3 37.0 101.0
-------------------------------------------------------------------
Please send questions or suggestions about this web page to beatnic@aset.psu.edu
ASET | ITS | Penn State
|
