Unexpected Behavior or Bug in PROC HPBIN

NguyenThanhTra — Fri, 23 May 2025 13:14:56 GMT

I encountered a rather odd issue when using PROC HPBIN that I’d like to share.

First, I used PROC HPBIN to perform binning on a dataset TEMP01and saved the resulting binning information into a mapping table (TEMP_MAPPING)

DATA TEMP01;
	LENGTH ID 8;

	DO ID=1 TO 1000;
		X1 = RANUNI(101);
		X2 = 10*RANUNI(201);
		X3 = 100*RANUNI(301);
		OUTPUT;
	END;
RUN;

PROC HPBIN DATA=TEMP01 OUTPUT=BIN_TEMP01 NUMBIN=10 BUCKET;
	INPUT X1-X3;
	ODS OUTPUT MAPPING=TEMP_MAPPING;
RUN;

This part worked as expected.

However, when I tried to apply the same binning to a different dataset (TEMP02) using the saved mapping, the process failed — but only when the variable order in the new dataset was different from the original

PROC SQL NOPRINT;
	CREATE TABLE TEMP02 AS SELECT X3, X2, X1 FROM TEMP01;
QUIT;

PROC HPBIN DATA=TEMP02 BINS_META=TEMP_MAPPING OUTPUT=BIN_TEMP02 NUMBIN=10 BUCKET;
	INPUT X1-X3;
RUN;

The result:

It also makes the issue very difficult to detect and debug. I believe this behavior should be addressed, or at the very least clearly documented, as it can lead to significant confusion and incorrect processing.

Re: Unexpected Behavior or Bug in PROC HPBIN

SASKiwi — Sun, 25 May 2025 20:44:13 GMT

If you want an official response to your issue then raising a Tech Support track is the best option.

topic Unexpected Behavior or Bug in PROC HPBIN in Statistical Procedures

Unexpected Behavior or Bug in PROC HPBIN

Re: Unexpected Behavior or Bug in PROC HPBIN