When using "parceling" as the Reject Inference Method for the Reject Inference node in SAS^® Enterprise Miner™, if at least one reject observation is inferred within each bin, then the "rejects population" is duplicated. As a result, there will be an inaccurate number of observations in the augmented data set.

No error or warning messages will be issued.

To determine whether or not the problem occurred, compare the total number of observations exported from the Reject Inference node to number of observations imported by the Reject Inference node. If the number of observations exported from the Reject Inference node is equal to the number of observations from the Scorecard node plus twice the number of observations in the rejects data set, then the problem occurred.

If there is at least one bin with no inferred reject observation, then the problem does not occur. You might be able to avoid the problem by using "Manual" for the Score Range Method and trying to get the _inferred_event_count to equal zero for at least one bin.

Select the Hot Fix tab in this note to access the hot fix for this issue.

Operating System and Release Information

Product Family	Product	System	Product Release		SAS Release
			Reported	Fixed*	Reported	Fixed*
SAS System	SAS Enterprise Miner	Microsoft Windows XP 64-bit Edition	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Microsoft Windows Server 2003 Enterprise 64-bit Edition	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Microsoft Windows Server 2003 Datacenter 64-bit Edition	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Microsoft® Windows® for 64-Bit Itanium-based Systems	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Microsoft Windows 2000 Advanced Server	5.3		9.1 TS1M3 SP4
		Microsoft Windows 2000 Datacenter Server	5.3		9.1 TS1M3 SP4
		Microsoft Windows 2000 Server	5.3		9.1 TS1M3 SP4
		Microsoft Windows 2000 Professional	5.3		9.1 TS1M3 SP4
		Microsoft Windows Server 2003 Datacenter Edition	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Microsoft Windows Server 2003 Enterprise Edition	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Microsoft Windows Server 2003 Standard Edition	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Microsoft Windows XP Professional	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		64-bit Enabled AIX	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		64-bit Enabled HP-UX	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		64-bit Enabled Solaris	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		HP-UX IPF	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Linux	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Linux on Itanium	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Solaris for x64	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2
		Tru64 UNIX	5.3	6.1_M2	9.1 TS1M3 SP4	9.2 TS2M2

* For software releases that are not yet generally available, the Fixed Release is the software release in which the problem is planned to be fixed.

---

A fix for this issue for SAS Credit Scoring for Enterprise Miner 5.3 is available at:

http://www.sas.com/techsup/download/hotfix/creditscr53.html#036517

---

When using parceling as a reject inference method, if at least one reject observation is inferred within each bin, then the rejects population is duplicated resulting in an inaccurate number of observations in the augmented data set.

Type:	Problem Note
Priority:	alert

Date Modified:	2010-07-09 10:21:03
Date Created:	2009-07-13 14:00:28