The HPBIN Procedure

This example shows quantile binning that is executed in distributed mode. Most of this example is the same as the pseudo–quantile binning example (see Example 4.2: Pseudo–Quantile Binning in Distributed Mode), so you can easily compare these two binning methods. The following DATA step generates 1,000,000 observations:

data ex12;
   length id 8;
   do id=1 to 1000000;
      x1 = ranuni(101);
      x2 = 10*ranuni(201);
      output;
   end;
run;    

You can run PROC HPBIN in distributed mode by specifying valid values for the NODES=, INSTALL=, and HOST= options in the PERFORMANCE statement. An alternative to specifying the INSTALL= and HOST= options in the PERFORMANCE statement is to set appropriate values for the GRIDHOST and GRIDINSTALLLOC environment variables by using OPTIONS SET commands. See the section Processing Modes in Chapter 3: Shared Concepts and Topics, for details about setting these options or environment variables.

The following statements provide an example. To run these statements successfully, you need to set the macro variables GRIDHOST and GRIDINSTALLLOC to resolve to appropriate values, or you can replace the references to macro variables with appropriate values.

ods output BinInfo=bininfo;
ods output PerformanceInfo=perfInfo;
ods output Mapping=mapTable;
ods listing close;
proc hpbin data=ex12 output=out numbin=10 quantile;
   input x1-x2;
   performance nodes=4 nthreads=8
   host="&GRIDHOST" install="&GRIDINSTALLLOC";
run;
ods listing;

The "Performance Information" table in Output 4.3.1 shows the grid setting.

The "Binning Information" table in Output 4.3.2 shows the binning method, number of bins, and number of variables.

The "Mapping" table in Output 4.3.3 shows the level mapping of the input variables. As you can see from this table, when the binning method is quantile, PROC HPBIN assigns the same number of observations to each bin for the input variables if possible.