To Lecture Notes

proc sort

• proc sort sorts a dataset in order of the sort variable.
   
• The dataset can be sorted in descending order like this:
  

  proc sort;
     by descending gpa;
  

• The default collating sequence for x/OS (IBM mainframes) is EBCDIC.
   
• The default collating sequence for most other operating systems, for example Windows, is ASCII.
   
• You can change the collating sequence to EBCDIC like this:
  

  proc sort sortseq=EBCDIC;
  

• Here are some other sortseq options.
   
• Try sorting these names with proc sort:
  

  Zoe anna Jason brandon
  

• Change the sort order to ignore case:
  

  proc sort sortorder=linguistic 
            (strength=primary);
     by name;
  

• Next try sorting these character values:
  

  1A 12B 2R 122P 10Z
  

• Change the sort order to sort in numeric order:
  

  proc sort sortorder=linguistic 
            (numeric_collation=on);
     by name;

• See the ProcSort Example.
   
• The nodup option removes duplicate observations from the dataset:
  

  proc sort data=kids out=sortedkids nodup;
     by gender;
  

• The nodupkey option removes observations with duplicate by variables:
  

  proc sort data=kids out=sortedkids nodupkey;
     by gender;
  

• The threads (nothreads) options allow (prevent) multiple threads to be used when sorting when it makes sense. For example, multithreaded sorting does not make sense on a machine with only one cpu.
   
• More details are given about proc sort options are discussed in the following reference.
   
• Reference: Base SAS(R) 9.2 Procedures Guide
  PROC SORT Statement
  http://support.sas.com/documentation/cdl/en/proc/61895/HTML/default/viewer.htm#a000146878.htm