|
|||
|
Hi.
I was running a tree model in Enterprise Miner with standard defaults. I had some character variables enter the model. They had been classified as nominal, but should have been ordinal, so I reclassified them as ordinal in a data set attributes node, and ran a new tree on the modified data. Now, the tree fails unless I remove those variables from the model. The failure always occurs in PROC SPLIT as a "SPLIT internal error. Should not happen." Then it gives some traceback information and says "Read Access Violation in Task ( SASDSVX ] Exception occurred at (67E6D9C8) Task Traceback". Has anyone here run into something similar? As I alluded above, it appears to be related to ordinalizing character variables. The tree model will run fine if I exclude the ordinal variables, and it ran fine before I changed their attributes from nominal to ordinal. -- TMK -- "The Macro Klutz" |
|
|
||||
|
||||
|
|
![]() |
| Thread Tools | |
| Display Modes | |
|
|
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| SV: Enterprise Guide or Miner on Red Hat Linux? | Fredrik Hansson | Newsgroup comp.soft-sys.sas | 0 | 04-15-2008 07:03 PM |
| Re: SAS Enterprise Miner and Text Miner | Frank Boggasch | Newsgroup comp.soft-sys.sas | 0 | 07-31-2007 07:36 AM |
| Re: NEED SAS examples on Conjoint Analysis and MDS | Wu Consulting | Newsgroup comp.soft-sys.sas | 0 | 10-23-2006 01:29 PM |
| Re: Enterprise miner problem | toby dunn | Newsgroup comp.soft-sys.sas | 0 | 09-06-2006 06:05 PM |
| Re: What's the algorithm used by SAS Enterprise Miner 5.1 to create a entropy tree? | Vadim Pliner | Newsgroup comp.soft-sys.sas | 0 | 10-25-2004 03:48 PM |