DQSCHEME Procedure

Example 2: Creating Schemes

The following example generates three schemes in SAS format. Note that the locale ENUSA is assumed to have been loaded into memory as part of the locale list.

/* Create the input data set. */
data vendors;
  input city $char17. state $char22. company $char36.;
datalines;
Detroit          MI                          Ford Motor
Dallas           Texas                       Wal-mart Inc.
Washington       District of Columbia        Federal Reserve Bank

/* See Example 1: Creating an Analysis Data Set for the full data set. */

Washington District of Columbia   Federal Reserve Bank
Atlanta          GEORGIA                     Target
;
run;

proc dqscheme data=vendors nobfd;
  create matchdef='City (Scheme Build)' var=city
    scheme=city_scheme locale='ENUSA';
  create matchdef='State (Scheme Build)' var=state
    scheme=state_scheme locale='ENUSA';
  create matchdef='Organization (Scheme Build)'
    var=company scheme=org_scheme locale='ENUSA';
run;

title 'City scheme';
proc print data=work.city_scheme;
run;

title 'State scheme';
proc print data=work.state_scheme;
run;

title 'Organization scheme';
proc print data=work.org_scheme;
run;

Details

Notice that this example did not create and immediately apply one or more schemes within the same step. After you create schemes, it is important that someone familiar with the data review the results. In this particular example, the City scheme chose Dalas as the transformation value for the city of Dallas. Although the values Dalas and Dallas were correctly clustered, you would probably prefer Dallas to be the transformation value.

Note: This example is available in the SAS Sample Library under the name DQSASSCH.