Example of groupformat option on the by statement in a data step.

Author: ChrisRiddiough

By Syntax/FORMAT Procedure/Using the By Groupformat Statement to Process Data.sas

@@ -0,0 +1,89 @@
+/************************************************************************************************
+ USING THE BY GROUPFORMAT STATEMENT TO PROCESS DATA                                                                                   
+     This program creates a format that can be used to group data. It shows the use of the groupformat option in the data step.                                                                         
+     Keywords: PROC FORMAT, FORMAT, VALUE, GROUPFORMAT                                                               
+     SAS Versions: SAS 9, SAS Viya                                                    
+     Documentation: https://documentation.sas.com/?cdcId=pgmsascdc&cdcVersion=default&docsetId=proc&docsetTarget=p1xidhqypi0fnwn1if8opjpqpbmn.htm
+     1. Create a data set on which to test the new format. The data is sorted by the value to be formatted.
+     2. Use PROC FORMAT with a value statement. Provide ranges for which you want to assign different grouping levels. 
+     3. Use the groupformat option on the by statement to allow the data step to read the data by groups.  
+     4. Because the data is being read with a by statement the first. and last. variables are created. In this example, the
+        first.salary will be the lowest salary observation for a group and the last.salary observation will be the highest.
+     5. In the PROC REPORT apply the format to see the results.
+************************************************************************************************/
+data payroll; /*1*/
+   infile cards;
+   input EmployeeID Salary BirthDate : date9. HireDate : date9.;
+   cards;
+120185 25080 07APR1978 01DEC2010 
+121101 25390 28AUG1990 01NOV2010 
+121092 25680 08MAR1978 01AUG2006 
+121106 25880 02FEB1973 01FEB2000 
+121072 26105 10JAN1983 01SEP2008 
+120178 26165 23NOV1958 01APR1978 
+121027 26165 05MAY1968 01DEC1993 
+121096 26335 18MAY1973 01MAY1999 
+120112 26550 17FEB1973 01JUL1994 
+121002 26650 18SEP1958 01DEC1983 
+120687 26800 25MAY1983 01AUG2008 
+120183 26910 10SEP1973 01DEC2010 
+120999 27215 28DEC1963 01APR1988 
+120153 27260 07MAY1983 01JAN2002 
+121059 27425 25OCT1963 01APR1985 
+120164 27450 26NOV1963 01FEB1986 
+120120 27645 05MAY1948 01JAN1978 
+121139 27700 19AUG1963 01JUL1991 
+121075 28395 23DEC1948 01JAN1978 
+120715 28535 12JUN1983 01FEB2009 
+121042 28845 04APR1983 01NOV2003 
+120799 29070 23MAR1983 01NOV2002 
+121105 29545 09MAY1983 01JAN2007 
+121061 29815 16JUL1958 01JUL1988 
+121062 30305 28OCT1988 01AUG2010 
+121019 31320 25JUN1990 01JUN2008 
+121032 31335 24FEB1992 01MAR2010 
+121080 32235 24JAN1963 01SEP1991 
+121099 32725 19MAR1983 01MAY2004 
+120787 34000 22AUG1973 01JAN2000 
+120731 34150 15MAR1963 01SEP1987 
+120754 34760 02JUN1992 01MAY2010 
+120683 36315 12NOV1958 01JAN1978 
+120757 38545 18MAR1948 01JAN1978 
+120789 39330 14JUL1968 01DEC1986 
+120774 45155 17SEP1982 01MAR2006 
+120796 47030 12MAY1958 01MAR1987 
+120793 47155 08AUG1973 01JUN2000 
+120668 47785 23OCT1953 01DEC1982 
+120813 50865 14SEP1973 01JAN1997 
+120804 55400 11FEB1948 01JAN1978 
+120712 63640 12JUN1953 01JAN1978 
+121145 84260 22NOV1953 01APR1980 
+120260 207885 02DEC1968 01NOV1988 
+120259 433800 25JAN1968 01SEP1993 
+;
+run;
+
+proc format; /*2*/
+   value salgrp low-<30000='Under $30,000'
+                30000-<35000='$30,000 to $35,000'
+				35000-<50000='$35,000 to $50,000'
+				50000-high='Over $50,000';
+run;
+
+data highlowsal;
+   set payroll;
+   by groupformat salary; /*3*/
+   format salary salgrp.;
+   if first.salary then output; /*4*/
+   if last.salary then output; 
+run;
+
+proc report data=highlowsal nowd; /*5*/
+   columns EmployeeID Salary Salary=SalGrp HireDate;
+   define salary / display format=dollar12.;
+   define hiredate / format=date9.;
+   define salgrp / format=salgrp. 'Salary Group';
+   title 'Lowest and Highest Salary by Salary Group';
+run;
+
+title;