libname u "U:\";
data u.incomes;
input gender age income;
format income dollar8.;
datalines;
0 50 60000
1 45 80000
1 30 25000
0 25 18000
1 72 40000
;
run;
Using and Saving Custom Formats
SAS enables the use of value labels, called formats (e.g. 0 is male, 1 is female) by allowing you to define custom formats. A difficulty is that these formats are not saved as part of the data set they label.
SAS Formats
As an example, consider a data set of individuals. For each individual you have their gender, their age, and their income. We want to do three things with this data:
- read it in and prepare it for analysis,
- get basic summary statistics, and
- regress income on the other variables.
You might put each step in a separate SAS program. However you want to apply the same value labels to gender in all three programs, so you need a way to share a custom format between programs.
Start by reading in the example data. In this example we'll put the data in our script, using DATALINES
.
The data is saved as a permanent data set which we can use in later programs without re-reading the text data. In addition, we have assigned income
a format, DOLLAR8.
. This is one of many formats supplied by SAS. It adds US dollar symbols and commas to the display of income
data values, using up to 8 characters.
proc freq data=u.incomes;
table income;
run;
The FREQ Procedure
Cumulative Cumulative
income Frequency Percent Frequency Percent
-------------------------------------------------------------
$18,000 1 20.00 1 20.00
$25,000 1 20.00 2 40.00
$40,000 1 20.00 3 60.00
$60,000 1 20.00 4 80.00
$80,000 1 20.00 5 100.00
But consider a frequency table for gender
, which we have not given any value labels.
proc freq data=u.incomes;
tables gender;
run;
The FREQ Procedure
Cumulative Cumulative
gender Frequency Percent Frequency Percent
-----------------------------------------------------------
0 2 40.00 2 40.00
1 3 60.00 5 100.00
Similarly, run a regression (just some of the output is shown here).
ods select parameterestimates;
proc glm data=u.incomes;
class gender;
model income = gender age / solution;
run;
The GLM Procedure
Dependent Variable: income
Standard
Parameter Estimate Error t Value Pr > |t|
Intercept 22194.63822 B 49787.49674 0.45 0.6994
gender 0 -3198.74162 B 31834.20993 -0.10 0.9291
gender 1 0.00000 B . . .
age 533.44276 939.69343 0.57 0.6275
It would be easier to interpret the results of these statistical procedures if there were value labels for gender
, rather than having to remember what “0” and “1” mean - it is not at all clear what these values mean here.
Defining and Using Formats
Formats in SAS are defined using PROC FORMAT
, and are applied to variables using a FORMAT
statement. So to apply value labels to the gender
variable, the first step is to define a format that associates 0 with male and 1 with female. We'll call this format “genderformat”.
Defining a Format
The key element of PROC FORMAT is a VALUE
statement, taking the form
VALUE formatname valuerange1=label1 valuerange2=label2 ...;
See VALUE statement for more details and options.
In the typical case there is a one-to-one correspondence between values and labels. (But see Using Formats to Collapse Categories to see how value ranges can be used to recode variables.)
proc format;
value genderformat
0= 'male'
1= 'female'
;
run;
Using a Format
Next you need to associate that format with the gender
variable, using a FORMAT statement.
format gender genderformat.;
Note that when a format is defined the name does not include a period, e.g.”genderformat”. When a format is used the name does include a period, e.g. “genderformat.”.
This statement could appear in either a DATA
step or a PROC
step. Assigned in a DATA
step, the format is used in all subsequent PROC
s that process that data set. Assigned in a PROC
step, it is used for only that specific step.
For example, the frequency table would be easier to read if we produced it like this:
proc freq data=u.incomes;
format gender genderformat.;
tables gender;
run;
The FREQ Procedure
Cumulative Cumulative
gender Frequency Percent Frequency Percent
-----------------------------------------------------------
male 2 40.00 2 40.00
female 3 60.00 5 100.00
In this case, the data set is unformatted, and the value labels are assigned only to produce the one table.
Reusing Formats
A difficulty is that genderformat
is deleted when the SAS session that defined it ends. How can we save it so that we can use it in later SAS sessions?
There are three solutions that might occur to you.
- Include the
PROC FORMAT
in every SAS script where you will use it. - Put the
PROC FORMAT
in a separate script, and call that script from all the other scripts that will use it. - Save the formats in a SAS catalog file, and include a reference to the catalog in other scripts.
Include the Format Definitions in All Your Programs
Simply copy-and-paste the PROC FORMAT into every SAS program file that will use these value labels.
This option might be good if you only have a few scripts that need to use the same set of formats. However, if you ever revise your value labels, you should make the revision in each of your scripts.
Use a Separate Formats Script
The difficulty of keeping numerous copies of a PROC FORMAT, saved in numerous separate files, in sync motivates the second option: reusing a common “formats script” in multiple files.
Put all of the formats in a separate file (with any name)
----- formats.sas -----
proc format;
value genderformat
0= 'male'
1= 'female'
;
run;
-----------------------
Then programs that rely on these formats can %INCLUDE%
this file.
----- procfreq.sas -----
%include% "formats.sas";
proc freq data=u.incomes;
format gender genderformat.;
tables gender;
run;
------------------------
Like the first option, the formats are recreated each time they are reused.
Saving Formats in a Catalog
The third option is more efficient, saving the formats that have been created. While this will not appreciably speed up most scripts, it does make the log much cleaner by avoiding all the PROC FORMAT output, making it easier to focus on more important messages in the log.
To understand how this works, consider the messages that PROC FORMAT writes to your log.
2 proc format;
3 value genderformat
4 0= 'male'
5 1= 'female'
6 ;
NOTE: Format GENDERFORMAT has been output.
7 run;
The format definition is “output” (saved) in a file named FORMATS.SAS7BCAT, located in your WORK library. This is the default location for saving user formats, a file that is separate from your data file.
In order to save format catalog, you'll add a LIBRARY
option to the PROC FORMAT statement, pointing to a permanent library.
When you need to use a format, SAS will automatically look for a formats catalog, first in your WORK library and then in a library named “LIBRARY” (if it has been defined).
(Where SAS searches for format catalogs is an option which can be configured with the FMTSEARCH
system option.)
A typical approach to defining a library named LIBRARY would be to point LIBRARY to the same location where the data set will be stored: both the data set and the formats catalog are saved side-by-side.
libname u "U:\";
libname library (u);
proc format library=library;
value genderformat
0= 'male'
1= 'female'
;
run;
data u.incomes;
input gender age income;
format income dollar8. gender genderformat.;
datalines;
0 50 60000
1 45 80000
1 30 25000
0 25 18000
1 72 40000
;
run;
In the preceding code:
- libname LIBRARY refers to wherever U refers to
- PROC FORMAT saves formats in the catalog file LIBRARY.FORMATS
- the variable GENDER is assigned value labels from GENDERFORMAT in the DATA step.
Now separate programs can use this variable as long as SAS can find the format. In other words, programs which use this data set must include a LIBNAME LIBRARY
definition.
libname u "U:/";
libname library (u);
ods select parameterestimates;
proc glm data=u.incomes;
class gender;
model income = gender age / solution;
run;
The GLM Procedure
Dependent Variable: income
Standard
Parameter Estimate Error t Value Pr > |t|
Intercept 22194.63822 B 49787.49674 0.45 0.6994
gender 0 -3198.74162 B 31834.20993 -0.10 0.9291
gender 1 0.00000 B . . .
age 533.44276 939.69343 0.57 0.6275
Lost Catalogs
Data sets include information about what format names have been assigned to variables, but the data set does not contain the actual formats. As we have seen, these are stored in a separate file. This is useful when we want to use the same formats with multiple data sets. But it also means that data sets and format catalogs are easily separated (like a little kid on a large family outing).
If you ever try to work with a data set that has been formatted using formats you don't have access to, the following FORMAT statement can be used to tell SAS to strip selected variable formats. This can be used in either a DATA step or a PROC step.
FORMAT varlist;
Because this FORMAT statement does not name any format, any variables named in the variable list are “assigned” a “null” format.
A particularly useful variable list is the special keyword _ALL_
. To remove all format information from a data set, make a copy of the data, assigning null formats to all variables. (This also strips SAS formats, such at date, time, and currency formats.)
data incomes;
set u.incomes;
format _ALL_;
run;
Last Revised: 8/14/2024