1. Copyright © 2004, SAS Institute Inc. All rights reserved. SASHELP Datasets A real life example Barb Crowther SAS Consultant October 22, 2004

2. Copyright © 2004, SAS Institute Inc. All rights reserved. 2 Situation  The Canadian government requires all financial institutions to report suspicious financial transactions, including: Cash Money Orders Casino Chips Real Estate ….

3. Copyright © 2004, SAS Institute Inc. All rights reserved. 3 Situation – Data  The data is extracted from many different applications and saved to a flat file as a ‘report’.  Excluding the header and subheader, there were 9 different parts for each report  Data formatting was very specific and varied from part to part  The number of fields varied from 4 to 35,  Field lengths varied between 1 and 400  The part lengths varied between 139 and 507 characters.

4. Copyright © 2004, SAS Institute Inc. All rights reserved. 4 Types of Required Information ……

5. Copyright © 2004, SAS Institute Inc. All rights reserved. 5 The first kicker  Not all reports needed all parts.  Some parts were always required, others only some of the time.  Which parts were included was dependant on the data.  But if the part was required, it had to be perfect!!!

6. Copyright © 2004, SAS Institute Inc. All rights reserved. 6 The second kicker  The users wanted to be able to edit each part before it was sent to the government – but because of the tool they used, they could not insert (or delete) a missing part.  So even all the fields were missing from the source data, the part had to be included

7. Copyright © 2004, SAS Institute Inc. All rights reserved. 7 So……….  The application had to insert each required part  The only information I would get is a sequence number.

8. Copyright © 2004, SAS Institute Inc. All rights reserved. 8 Attempt #1 – Hard code the data step For each of the nine parts….. …. For each of the 10 to 35 fields per part…. …I could write a “length variable ($) n;” statement. Oh, and by the way, did I tell you that the government regularly changes part content?

9. Copyright © 2004, SAS Institute Inc. All rights reserved. 9 Attempt #2 – Investigate system options Options obs=0; Data parta1; set input.parta1; Run; Options obs=max; The problem with this code is the output dataset has no observations and I needed one, even if there was no data.

10. Copyright © 2004, SAS Institute Inc. All rights reserved. 10 Attempt #3 – Look at SASHELP.datasets  The SASHELP datasets contain information about the current SAS session including all the members of all the libraries (SASHELP.VMEMBER) all the columns of each member (SASHELP.VCOLUMN)

11. Copyright © 2004, SAS Institute Inc. All rights reserved. 11 VCOLUMN Contents Variable NameVariable LabelCreate Date … FormatColumn Format InformatColumn Informat LabelColumn Label LengthColumn Length LibnameLibrary Name MemnameMember Name TypeColumn Type …

12. Copyright © 2004, SAS Institute Inc. All rights reserved. 12 Accessing V* Tables  Accessing the V* Tables can be done using PROC, SQL, or Data statements proc print data=sashelp.vtable; where libname='WORK'; run; proc sql; create view work.options as select * from dictionary.options;

13. Copyright © 2004, SAS Institute Inc. All rights reserved. 13 So how does this help me?  Step 1: Get a list of all the variables (and their attributes) required for the “empty dataset”.  Step 2: Move all that information into macro variables  Step 3: Create a dataset template  Step 4: Create the empty dataset.

14. Copyright © 2004, SAS Institute Inc. All rights reserved. 14 Step 1: Variables and their attributes proc sql noprint; create table &table._vars as select name, type, format, length, label from sashelp.vcolumns where upcase(libname)=upcase("&inset") and upcase(memname) = upcase("&table") ; quit; For this example, our inset will be Work and our table Txns.

15. Copyright © 2004, SAS Institute Inc. All rights reserved. 15 Step 1: Work.Txns_vars VCOLUMNS Output NameTypeFormatLengthLabel Tran_dateChar8 Tran_Post_DateChar8 Tran_CurrencyChar$3.003Currency_code Tran_timeChar4 Tran_AmountNum8 Teller_idChar$15 …

16. Copyright © 2004, SAS Institute Inc. All rights reserved. 16 Step 2: Create macro variables using Txns_vars from Step 1 data _null_; set &table._vars end=eof; call symput('var'||left(put(_n_,3.)),name); if format ne ' ' then call symput('fmt'||left(put(_n_,3.)),format); else if upcase(type) = 'CHAR' then call symput('fmt'||left(put(_n_,3.)),'$'|| put(length,3.)||'.'); if label ne ' ' then call symput('label'||left(put(_n_,3.)),label); if eof then call symput('var_cnt',left(put(_n_,3.))) ; run;

17. Copyright © 2004, SAS Institute Inc. All rights reserved. 17 Step 2: Macro Output from the SASLOG

18. Copyright © 2004, SAS Institute Inc. All rights reserved. 18 Step 3: Create the dataset template Work.Txn_tpl using the Step 2 macro variables %let i = 1; data &table._tpl; %do i=1 %to &var_cnt; attrib &&var&i format= &&fmt&i label = "&&label&i"; %end;

19. Copyright © 2004, SAS Institute Inc. All rights reserved. 19 Step 3: Create the dataset template generated code to create Work.Txn_tpl Value of iGenerated Code 1Attrib transaction_key format = 20. label = ‘transaction_key”; 2Attrib transaction currency format = $3. label=‘Currency Code’; …

20. Copyright © 2004, SAS Institute Inc. All rights reserved. 20 Step 3: Create a dataset template Get a list of the required variables %global &table._var_list; proc sql noprint; select distinct name into :&table._var_list separated by ' ' from &table._vars ; quit; Results in a macro variable called txn_var_list with a value of TRANSACTION_KEY TRAN_CURRENCY …

21. Copyright © 2004, SAS Institute Inc. All rights reserved. 21 So where are we?  We have a report with a known sequence number, but no data  We know what variables are required &txn_var_list  We know the variables’ attributes &&var&i format= &&fmt&i label = "&&label&i";

22. Copyright © 2004, SAS Institute Inc. All rights reserved. 22 Step 4: Create the empty dataset

23. Copyright © 2004, SAS Institute Inc. All rights reserved. 23 Step 4: Code to generate the dataset data &table._miss_data; retain &&&table._var_list; set result (keep=seq_num); if _n_ = 1 then set &table._tpl(drop=seq_num); run;

24. Copyright © 2004, SAS Institute Inc. All rights reserved. 24 Thoughts…  Writing the macros took longer than hard coding the attribute statements.  But, if there are any future changes, I won’t have to do very much (if any).  The macros can be used in other applications…

25. Copyright © 2004, SAS Institute Inc. All rights reserved. 25 Suggested readings  The SASHELP Library: It Really Does Help You Manage Data by Melinda Thielbar http://support.sas.com/sassamples/bitsandbytes/sashel p.htmlhttp://support.sas.com/sassamples/bitsandbytes/sashel p.html  You Could Look It Up: An Introduction to SASHELP Dictionary Tables by Michael Davis You Could Look It Up: An Introduction to SASHELP Dictionary Tables http://www2.sas.com/proceedings/sugi26/p017-26.pdf

26. Copyright © 2004, SAS Institute Inc. All rights reserved. 26 Copyright © 2004, SAS Institute Inc. All rights reserved. 26