PICTURE Perfect: In depth look at the PICTURE format.
Carry W. Croghan, US-EPA, Research Triangle Park, NC.
EPA/600/A-04/088
ABSTRACT
SAS PICTURE format is a very powerful tool. The different options allow the display of data in a more picturesque
manner adding symbols, qualifiers, and comments to the data points without modifying the data. With a PICTURE
format one can create a series of templates for displaying the data. The PICTURE format even allows the user to
generate their own date and time formats. In addition to the options available with the VALUE formats, PICTURE
format has options such as DATATYPE, NOEDIT, PREFIX, MULTIPLIER, and FILL. With its multitude of options, the
PICTURE format is very impressive, but it can also be confusing. It is not always clear how different options will
interact. If not carefully applied, the PICTURE format can produce misleading information. This presentation will look
at the structure and the options available, removing some of the mystery of the PICTURE format and revealing a
powerful formatting tool.
INTRODUCTION
Formats allow you to display the data in a different structure without modifying the data. There are two general types of
formats that a user can define:
1.	VALUE - assigns a label or text string to a range of values,
2.	PICTURE - creates templates in which you define how the numbers are displayed. With a PICTURE format
you can output text strings as well as the numeric values.
PROC FORMAT SYNTAX
To define a format you use the PROC FORMAT procedure. The basic syntax is
proc format ;
picture name <(options)>
rangel = ^Templatel' <(options)>
template2' <(options)>
rangek = ^templatek' <(options)> >;
A name is required. It is an alphanumeric string of length up to 8 with the additional restriction that it cannot end with a
number. Also, it cannot be a name of an existing SAS defined format.
A range can be a single value (0), a set of values separated by commas (i.e. 2,4,6), or an interval of values (i.e. 1-6).
There are 3 key words that can be used in defining the range: Low - smallest non-missing value, High - largest value,
and Other- all values not in any of the ranges. To define an exclusive range, the symbol '<' is used. This can be used
at either the beginning or end of the range. For example, low -<0 would include all the negative values and would not
include 0.
A template consists of digit selectors and/or sequence of characters within quotes. The template must start with a digit
selector. The digits are interpreted as numeric place holders in which the data are placed. A zero is a place holder that
does not print leading zeros. All other digits (1-9) are place holders that do print leading zeros. It is tradition to only use
0 and 9 in the templates. This is a convention that all the examples will follow. There is a limit of 16 digit selectors in a
template. Most characters are message characters. They will be displayed exactly as you have written them. There
are some special characters combinations that are directives used in the creation of date and datetime templates.
These will be discussed in detail later on.
1

-------
This is a simple example of a PICTURE format for checking a validity of a month variable.
proc format;
picture mon
1-12 ='99'
other = ^Not a valid month';
run;
There are two range specifications and two templates. The '99' in the first template dictate that a leading zero is
printed. The second template is used for all values outside of rangel This template is a text message.
Here is how some values would be displayed.
Original data
Formatted data
0
Not a valid month
1
01
2
02
13
Not a valid month
10
10
Changing the format slightly, the leading zeros can be removed by using 0 as the digit selector.
proc format;
picture mon
1-12 ='00'
other = ^Not a valid month';
run;
Original data
Formatted data
0
Not a valid month
1
1
2
2
13
Not a valid month
10
10
What happens if the "other" range is removed?
Original data
Formatted data
0
0
1
1
2
2
13
13
10
10
This is not a very effective format. The formatted data look no different from the unformatted data. The other range is
an important part of the format. Another way to format would be to use a VALUE format for the non-valid month values.
proc format;
value mon
low-
-------
Original data
Formatted data
1
01
2
02
13
13 is not valid
10
10
This format works fine for integer values. However, it would not work for non-integers.
To paraphrase the ancient Delphi Oracle, 'Know thy data'. This is especial important when working with PICTURE
formats. You need to know the magnitude and the precision of the data values. The following example uses the same
format with different data values.
Original data
Formatted data
-11.0
11 is not valid
1.5
01
30012.0
12 is not valid
Although none of the values are valid months, 1.5 is reported as valid, and for the others the print out is misleading.
The PICTURE format not only truncates the data to the right of the decimal, it truncates the data to left of the defined
template. It also takes the absolute value of the data.
It is important to check the data to verify the range and precision. The simplest method is to do a PROC FREQ. If it is
a large dataset with numerous distinct values which should be whole numbers, create an indicator variable to check if
the data are in fact whole numbers.
chk = (round(X)=X);
Then run a PROC MEANS with MIN and MAX options on the original variable X and the new indicator variable chk.
Taking a few minutes to check the data can save hours of later head scratching.
To understand how SAS applies a template, we need to examine the specific steps that SAS goes through. These
steps can be modified by many of the options; therefore the options are defined first and then the steps.
OPTIONS
There are several options available in PROC FORMAT. These options are divided into two general types:
1.	Overall options - apply to the entire format.
2.	Template options - apply to a specific range and template.
Most of the overall options can be used with VALUE as well as PICTURE formats.
Option
Description
DEFAULT=
Specifies a default length for the format.
FUZZ=
Specifies a fuzz factor for matching values to a range.
MAX=
Specifies a maximum length for the format
MIN=
Specifies a minimum length for the format
MULTLABEL
Allows your ranges to overlap. This option is only useful if the
called procedure allows multiple labels.
NOTSORTED
Keeps the data in the order that you defined.
ROUND
Round the data to the nearest integer before formatting.
3
-------
These template options are applicable only to PICTURE formats.
Option
Description
DATATYPE=
Specifies the type of format. Only needed if you are defining a date, time, or
datetime format. The only valid values are DATE, TIME, and DATETIME.
DECSEP
Specifies a decimal separator character; by default it is
DIG3SEP
Specifies the three digit separator character; by default it is
FILL=
Specifies a value to fill in the blank space.
MULTIPLIER^
MULT=
Specifies a value to multiply the data by before it is formatted.
NOEDIT
The template is printed as is, the numeric values are interpreted as part of the text
and data is not displayed.
PREFIX=
Specifies a string of characters to place before the formatted values.
HOW SAS APPLIES A TEMPLATE
In the first step, SAS locates the range that the data value is in. If you have used the MULTLABEL option, it locates the
primary range. What the primary range is depends on whether the NOTSORTED option is used. Without the
NOTSORTED option, the primary range is the one that is numerically first. To determine which range is numerically
first both the start and the end of the range is considered. The range 0-9 occurs before 0-99. The range 0-9999 occurs
before 10-999. With the NOTSORTED option, it is the first range that you define in which the data value is in. The
FUZZ factor is also applied at this step. The FUZZ factor is similar to rounding; however, it is only used to match a data
value to a range.
In the second step, SAS takes the absolute value of the data. SAS applies the multiplier in the third step. This value is
either implicit or explicit. If you used the MULT option, then the multiplier is explicitly defined. The implicit multiplier is
based on the template. It is 10n, where n is the number place to the right of the decimal indicator. (The decimal
indicator is usually You can used the DECSEP option to change it to another character.)
In the fourth step, unless the ROUND option is used the data value is truncated to fit in the template. If the ROUND
option is used the data value is rounded to the nearest integer value.
In the fifth step, the numeric data is converted to a character string. If the magnitude of the value is smaller than the
number of digit selectors in the PICTURE, SAS pads the values to the left with leading zeros.
Next the new character string is placed in the template starting with the right most digit selector. If the first digit selector
is a zero, the leading zeros are replaced with blanks or the alternative FILL value.
The final step is the placing of the prefix if the PREFIX option was used. If the template is not wide enough to
accommodate the data and the prefix, SAS will drop the prefix.
The best way to explain both the steps and the options is by example.
Say we have the following recovery fractions data values: 1.22849, 2.09283, 0.64999,
-0.09871, and 0.23302. Let us assume that a recovery between 65% and 175% is considered to be good, anything
larger than 175% is considered to indicate contamination, smaller than 65% indicates a bad recovery. Since recoveries
are calculated as absolute differences, a negative value indicates erroneous data. The data precision should only be at
the 0.001 level. So we want the data rounded and the ranges to be assigned to the appropriate precision. The
researchers prefer to look at the results as percent recovery.
The format statements would be.
4
-------
/*!*/
1*2*1
1*3*1
/* A*/;
STEP 1. SAS ASSIGNS A RANGE.
Original values
Range
1.22849
3
2.09283
4
0.64999
3 (FUZZ = 0.005)
-0.09871
1
0.23302
2
STEP 2. SAS TAKES THE ABSOLUTE VALUES.
Original values
Range
Absolute values
1.22849
3
1.22849
2.09283
4
2.09283
0.64999
3
0.64999
-0.09871
1
0.09871
0.23302
2
0.23302
STEP 3. SAS MULTIPLIES BY THE MULT VALUE.
Original values
Range
Absolute values
MULT
1.22849
3
1.22849
1228.49
2.09283
4
2.09283
2092.83
0.64999
3
0.64999
649.99
-0.09871
1
0.09871
98.71
0.23302
2
0.23302
233.02
STEP 4. SAS ROUNDS OR TRUNCATIONS.
Original values
Range
Absolute values
MULT
ROUND
1.22849
3
1.22849
1228.49
1228
2.09283
4
2.09283
2092.83
2093
0.64999
3
0.64999
649.99
650
-0.09871
1
0.09871
98.71
99
0.23302
2
0.23302
233.02
233
STEP 5. SAS CONVERTS TO CHARACTER AND PADS WITH LEADING ZEROS
Original values
Range
Absolute values
MULT
ROUND
Char
1.22849
3
1.22849
1228.49
1228
1228
2.09283
4
2.09283
2092.83
2093
2093
0.64999
3
0.64999
649.99
650
0650
-0.09871
1
0.09871
98.71
99
0099
0.23302
2
0.23302
233.02
233
0233
proc format;
picture recovery (round fuzz=0.005 default = 25)
low-<0 = '999.9% Not valid Data' (prefix='-' mult = 1000)
0-0.64 = '009.9% Bad recovery' (mult = 1000)
0.65 - 1.75 = '009.9%' (mult = 1000)
1.75<-high = '00009.9%' (prefix = 'Contamination. ' mult=1000)
5
-------
STEP 6. SAS REPLACES DIGIT SELECTORS IN THE TEMPLATE, KEEPING ONLY THE SPECIFIED LEADING ZEROS.
Original values
Range
Char.
Templates
Insert Data
1.22849
3
1228
'009.9%'
122.8%
2.09283
4
2093
'00009.9%'
209.3%
0.64999
3
0650
'009.9%'
65.0%
-0.09871
1
0099
'999.9% Not valid Data'
009.9% Not valid Data
0.23302
2
0233
'009.9% Bad recovery'
23.3% Bad recovery
STEP 7. SAS ADDS THE PREFIXES.
Original values
Range
Insert Data
Add Prefixes
1.22849
3
122.8%
122.8%
2.09283
4
209.3%
Contamination. 209.3%
0.64999
3
65.0%
65.0%
-0.09871
1
009.9% Not valid Data
-009.9% Not valid Data
0.23302
2
23.3% Bad recovery
23.3% Bad recovery
EXAMPLES OF SELECT OPTIONS
PREFIX AND ROUND
Since a template must start with a digit selector, the only way to get text in front of the data is to use the PREFIX option.
Here is an example of the PICTURE format using both the ROUND and PREFIX options. In this example we will be
formatting data for an accounting report. The position values will be reported in dollars and the negative values are in
()¦
proc format;
picture accA (round)
1qw-<0 = '0,000,009)'	(prefix = '(')
0-high = '0,000,009' (prefix = '$');
run;
Original data
Formatted data
-980.00
(980)
-745.60
(746)
200.00
$200
348.75
$349
880.20
$880
1420.00
$1,420
5000000.00
$5,000,000
If the ROUND option is omitted, the fractional portion of the data will be truncated.
proc format;
picture accB /*No round option */
low-<0 = '0,000,009)'	(prefix = '(')
0-high = '0,000,009' (prefix = '$');
run;
Original data
Formatted data
-980.00
(980)
-745.60
(745)
200.00
$200
6
-------
Original data
Formatted data
348.75
$348
880.20
$880
1420.00
$1,420
5000000.00
$5,000,000
Too small of a template as well as not using the ROUND
left.
proc format;
picture accB /*No round option */
low-<0 = '009)'
(prefix ='(')
0-high = '009'
(prefix = '$');
run;
option cause the data to be truncated both on the right and
Original data
Formatted data
-980.00
980)
-745.60
745)
200.00
$200
348.75
$348
880.20
$880
1420.00
$420
5000000.00
$0
Notice the lose of not only data information but also the prefix on the negative values.
NOEDIT
Another option that I frequently use is the NOEDIT option,
characters and not as digit selectors. For example,
proc format;
picture accB (round default=10
fuzz= .5)
low-<0 = '009)' (prefix = '(')
0-999 = '009' (prefix = '$')
1000-high = '> 1K' (noedit);
run;
It instructs SAS to treat any numbers within the template as
Original data
Formatted data
-980.00
(980)
-745.60
(746)
200.00
$200
348.75
$349
999.60
> 1K
1420.00
> 1K
5000000.00
> 1K
With DEFAULT option used, there is enough space for the '(' to be displayed for the negative values. The FUZZ factor
causes 999.6 to be assigned to the third range. Without FUZZ, the value 999.6 is not within any of the ranges.
(Remember, rounding is not done until step 4.) The NOEDIT indicates that the 1 in the last template should be printed
as is and not replaced with data values. Without the NOEDIT option, the output is misleading.
7
-------
Original data
Formatted data
-980.00
(980)
-745.60
(746)
200.00
$200
348.75
$349
999.60
OK
1420.00
OK
5000000.00
OK
CREATING YOUR OWN DATES FORMATS
SAS provides a plethora of date and time formats. If you want something more esoteric, you can create your own by
including special characters in the PICTURE template called DIRECTIVES. DIRECTIVES are only recognized by SAS
when you specify the option DATATYPE in the template. A DIRECTIVE is percent sign followed by a letter.
The permitted directives are
Directive
Description
%a
Abbreviated weekday name
%A
Full weekday name
%b
Abbreviated month name
%B
Full month name
%d
Day of the month as a number
(1-31)
%H
Hour (24-hour clock) as a number (0-23)
%l
Hour (12-hour clock) as a number (1-12)
%j
Day of the year as a number
(1-366)
%m
Month as a number (1-12)
%M
Minute as a number (0-59)
%p
either AM or PM
%S
Second as a number (0-59)
%U
Week number of the year (0,53)
%w
Weekday as a number (1= Sunday, 7)
%y
Year without century as a number (0-99)
%Y
Year with century as a number
%%
%
By default none of the DIRECTIVES generate leading zeros. By adding a 0 between the % and the letter of a numeric
directive, you can get leading zeros. For example, if you specify %0y, then for the year 2004 you will get 04.
With a DATATYPE PICTURE format, it is important to explicitly define the default length using the DEFAULT option.
SAS counts the space between the quotes as the default length unless otherwise directed. Since %B is only two
spaces it will never accommodate the full name of the month.
proc format;
picture mydate /*No Default */
low-high = '%B' (datatype=date);
run;
Original Data
(formatted with date9.)
Formatted data
14JUN2004
Ju
29JUL2004
Ju
8
-------
Original Data
(formatted with date9.)
Formatted data
17SEP2004
Se
When choosing a DEFAULT, take into account the largest possible value.
proc format;
picture mydate (default = 15)
low-high = '%B' (datatype=date);
run;
Original Data
(formatted with date9.)
Formatted data
14JUN2004
June
29JUL2004
July
17SEP2004
September
Here is a more elaborate date format.
proc format;
picture mydate (default = 30)
low-'04NOV2004'd = '%a %d of %B %Y' (datatype=date)
'04NOV2004'd - high = 'Sometime in the future';
run;
Original Data
(formatted with date9.)
Formatted data
09JUL2004
Fri 9 of July 2004
23AUG2004
Mon 23 of August 2004
070CT2004
Thu 7 of October 2004
21NOV2004
Sometime in the future
CONCLUSIONS
PICTURE formats allows the combination of text messages and data in a structured output. The steps that SAS goes
through to input the data into the template are important to remember. There are many options that must be used to
produce the desired output. Some important options are PREFIX, NOEDIT, ROUND, and DEFAULT. PREFIX allows
the placement of text before the numbers. This is important because a template must start with a digit selector. Any
text in front of the first digit selector is ignored by SAS. If any numbers are included in a text message, SAS will
interpret those as digit selectors unless the option NOEDIT is used. By default SAS truncates the data. The ROUND
option is important in order for the continuous data to not be truncated. Get all of the desired text output by explicitly
defining the length of the format by using the DEFAULT option. However, the most important thing to remember is to
"Know thy data". It is easy to produce misleading output with a PICTURE format if the template does not
accommodate the data, or the specified ranges do not capture all the data values. When used with care, a PICTURE
format is a powerful tool for displaying data.
REFERENCE
Lund, Pete, 2002 , "More than Just Value: A Look Into the Depths of PROC FORMAT". SUGI 27.
SAS, 1999, SAS Procedures Guide, Version 8. SAS Publishing.
CONTACT INFORMATION
Your comments and questions are valued and encouraged. Contact the author at:
Author Name: Carry Croghan
Company:	US - EPA
Address:	109 TW Alexander Drive
City state ZIP: RTP, NC 27709
9
-------
Work Phone: (919)5413184
Email:	croghan.carry@epa.gov
This paper has been reviewed in accordance with the United States Environmental Protection Agency's peer and administrative review
policies and approved for presentation and publication. Mention of trade names or commercial products does not constitute
endorsement or recommendation by EPA for use.
10
-------