Saving SAS Data Sets

Saving SAS Data

Save SAS data by either

  • using a permanent SAS library
  • using a quoted operating system file name

Introduction

In order for SAS to process any data, that data must be in the format of a SAS data set. This is a special SAS binary file which usually has the file extension “.sas7bdat”. These are typically saved by writing some SAS code, such as a DATA step.

Temporary and Permanent (Saved) Data Sets

Every SAS data set is stored in your computer’s file system, rather than being solely loaded in your computer’s memory. However, some data sets are temporary while others are permanent.

Temporary data are written to a temporary folder usually referred to as the WORK library, which is automatically deleted when SAS is shut down. Data sets referred to by a SAS name (more below) without a library reference are stored in the WORK library by default, so they are temporary.

SAS data sets may be created in several ways, and any of these may be created as either temporary or permanent.

  • a DATA step
  • many PROCs have “output data” options
  • all printable output tables can be saved as data using ODS

File References

In your SAS code, these data files can be addressed in either of two ways, either using an operating system name or using a SAS name. SAS names are usually preferred, because they are easier to read, write, and they make your code more portable.

SAS file names generally take the form

library-name.file-name

or just

file-name

The operating system file name might include the location (the directory or folder) as well as the file name. For example a Windows data file name might look like

"U:\SAS\class.sas7bdat"

while a Linux data file name could look like

"~/SAS/class.sas7bdat"

On either operating system the SAS name for these data files might look like

u.class

Operating system file names are ALWAYS enclosed in quotes, while SAS file names are NEVER in quotes. In operating systems names, the file path (location) is included as part of the file name specification. In SAS file names, the location is given by a prefix called a library name. In the SAS example above, u is a library name while class is the data file name.

SAS Names

The basic code to save a SAS data set using the SAS name is

libname z "U:/SAS";   /* Winstat example */
* libname z "~/SAS";  /* Linstat example */
data z.class;
  set sashelp.class;
run;

This produces a log which says that the file was successfully saved.

2          data z.class;
3            set sashelp.class;
4          run;

NOTE: There were 19 observations read from the data set SASHELP.CLASS.
NOTE: The data set Z.CLASS has 19 observations and 5 variables.

There are two key elements to this code. One is the line that says data z.class;, a "data statement", which instructs SAS where to write the file and what to name it. In this case, the file will be written wherever the library z points to, and will be named class.

The other key element, then, is the line that says libname z "~/SAS";, a "libname statement", which tells SAS that the library z points to the place the operating system calls "~/SAS” (on Linux, the folder within my home directory).

The data set name and the library name are chosen by you, but there are certain restrictions (see SAS names). SAS names typically begin with a letter or underscore, and are composed of letters and numerals and underscores - special characters like blanks or parentheses are not (usually) allowed. Library names may be up to eight characters, while data set names may be up to 32 characters.

(The set statement tells SAS where to get data from, and the run; statement tells SAS to execute the instructions accumulated so far.)

Operating System Names

Any place you can use a SAS name for a data set, you could instead use the operating system name. So the code that saves the data set, above, could have been written

*data "~/SAS/class.sas7bdat"; /* Linstat example */

data "U:/SAS/class.sas7bdat"; /* Winstat example */
  set sashelp.class;
run;

And again we’ll check the log.

2          *data "~/SAS/class.sas7bdat"; /* Linstat example */
3          
4          data "U:/SAS/class.sas7bdat"; /* Winstat example */
5            set sashelp.class;
6          run;

NOTE: There were 19 observations read from the data set SASHELP.CLASS.
NOTE: The data set U:/SAS/class.sas7bdat has 19 observations and 5 
      variables.

Pros and Cons

The advantages of using SAS names are

  • shorter file references, less clutter
  • if you move your project to a new folder or a new platform, all you have to change is the LIBNAME.

The advantages of using operating system names are

  • more flexibile, expressive, file names
  • no need for a libname statement

In our experience, most SAS programmers use SAS names.