SAS is a comprehensive statistical software that provides a wide variety of analytical capabilities, data management, and visualization options. It offers an interactive matrix programming language and integration with R. SAS is most often used for handling large-scale management systems, data mining, and statistical analysis.
Availability and Setup
SAS version 9.3 is the default version on the Shared Computing Cluster (SCC). If you wish to use a more recent version of SAS, you can do so using modules as explained here. Along with the Base SAS procedures, the current installation includes many packages. Listed below are a set of the most popular/useful packages:
- STAT – Statistical analysis
- GRAPH – High quality graphics for presentations
- ETS – Econometrics and Time Series Analysis
- FSP – Integrated tools for data entry, computation, query, editing, validation, display, and retrieval
- OR – Operations research
- AF – Applications facility (menus and interfaces)
- IML – Interactive matrix language
- QC – Quality control
- ACCESS – Transparently share data with non-native datasources
- CONNECT – Provides ability for SAS sessions on different platforms to communicate with each other
To enable the Graphics User Interface, SAS should be invoked under the X Window System on X terminals or workstations.
To run SAS interactively on the SCC, type:
To run SAS in command-line mode without the GUI, type:
scc1% sas -nodms
-nodms stands for No Display Manager. The system will invoke SAS system and it will give you a prompt with a line number and a question mark. To exit SAS in no display manager mode do:
To run a SAS program (e.g.
myprogram.sas) in batch (non-interactive) mode, type:
scc1% sas myprogram
scc1% sas myprogram &
if you wish to run it in the background.
When the program finishes there will be two files created in your working directory. One of them –
myprogram.log – contains the log of the SAS session; the other –
myprogram.lst contains the output from the program.
If you want to direct your output and/or log to other files, use the
scc1% sas myprogram -print myoutput -log mylog.log
In order to submit a batch or batch interactive job to the batch system for running on the compute nodes, use the
-l sas option when submitting your job, i.e.
scc1% qsh -l sas -pe omp 1
See http://www.bu.edu/tech/support/research/system-usage/running-jobs/ for more information about options when submitting a job.
Additional notes on default SAS options: On the login node (
scc1.bu.edu) the MEMSIZE is capped at 2GB and SORTSIZE is capped at 1GB. Threading is also turned off so a user cannot use more than one CPU when running interactively. These options are NOT CHANGEABLE by the user.
On the batch nodes in the SCC that SAS can run on the MEMSIZE is capped at 4GB and SORTSIZE is capped at 2GB. Threading is turned on but the default CPUCOUNT is set to 1. So by default a user will only use 1 CPU even if he or she asks for more slots. These options ARE CHANGEABLE by the user.
This allows you some flexibility in increasing the number of threads and memory if your job requires it but to do so you must run on one of the batch nodes.
The SAS manual shows the different ways a user can change these options.
To get help during an interactive GUI session select Help -> SAS Help & Documentation.
Most of the SAS documentation is also available online at http://support.sas.com/documentation/onlinedoc/index.html.
A catalog of textbooks available from SAS can be found online at http://support.sas.com/publishing/index.html.
We also have a simple SAS “hello world” example of our own.