Problems with Association Rules

Question

Hi all,
I am trying to use ODMr to run association rules and am getting the following error message:
Server task state: error
Server task detail:
ORA-40101: Data Mining System Error ORA-40101: Data Mining System Error ORA-40101: Data Mining System Error ODM_ASSOCIATION_MODEL-BUILD--20010
ORA-06512: at "SYS.DBMS_SYS_ERROR", line 105
ORA-06512: at "DMSYS.ODM_ASSOCIATION_RULE_MODEL", line 144
ORA-06512: at "DMSYS.ODM_ASSOCIATION_RULE_MODEL", line 1396
ORA-00902: invalid datatype
ORA-06512: at "SYS.DBMS_SYS_ERROR", line 105
ORA-06512: at "DMSYS.ODM_ASSOCIATION_RULE_MODEL", line 144
ORA-06512: at "DMSYS.ODM_ASSOCIATION_RULE_MODEL", line 1396
ORA-06512: at "DMSYS.DBMS_DATA_MINING", line 305
ORA-06512: at "DMSYS.DBMS_JDM_INTERNAL", line 157
ORA-06512: at "SYS.DBMS_SYS_ERROR", line 86
ORA-06512: at "DMSYS.DBMS_JDM_INTERNAL", line 179
ORA-06512: at line 1
This error is generated on the sample data as used in the ORACLE 10G Release 2 Data Mining Tutorial, and on other datasets that I tried.
My sense is that the algorithm is expecting data in the datatype of DM_NESTED_NUMERICALS, even though the documentation asks for transactional data. When I converted to nested columns, the prog errored out again as it was expecting a number rather than nested column.??

Answer 1

Hi Carolyn,
I reviewed the tutorial and tested out the example for Association Rules. The example uses SH.SALES and SH.PRODUCTS.
It worked fine.
ODMr converts the "transactional" data, in the case of Association Rules, into DM_NESTED_NUMERICALS.
When ODM documentation refers to transactional data, it implies the use of nested columns. This can be confusing given that older versions of ODM actually supported a "transactional format", which is now supported only through nested columns.
I copied the following view definition that was passed in to ODM to build the Association Rules model (based on the tutorial noted above). Take a look at them to get a better understanding on what ODMr is doing to prepare the data.
If you can give us a specific example including the tutorial table you used and the settings you specified in the Association Rules Wizard, I can review that to see what the problem might be.
Thanks, Mark
SELECT caseTable."DMR$CASE_ID", txnTable1."PROD_ID" FROM "DMUSER"."DM4J$VSALES657013432" caseTable, (SELECT "DMR$CASE_ID", CAST(COLLECT(DM_Nested_Numerical("NAME", "VALUE")) AS DM_Nested_Numericals) "PROD_ID" FROM "DMUSER"."DM4J$SALES423975876" GROUP BY "DMR$CASE_ID") txnTable1 WHERE caseTable."DMR$CASE_ID" = txnTable1."DMR$CASE_ID"
The view makes references to other tables/views defined below:
DM4J$VSALES657013432 is shown below:
SELECT distinct "DMR$CASE_ID" FROM
     (SELECT "CUST_ID","TIME_ID", DENSE_RANK() OVER (ORDER BY "CUST_ID","TIME_ID") DMR$CASE_ID FROM "SH"."SALES")
ORDER BY "DMR$CASE_ID"
DM4J$SALES423975876 is shown below:
SELECT DMR$CASE_ID, CAST("PROD_ID" AS VARCHAR2(30)) as NAME, 1 as VALUE
FROM
(SELECT "PROD_ID", "CUST_ID","TIME_ID", DENSE_RANK() OVER (ORDER BY "CUST_ID","TIME_ID") DMR$CASE_ID FROM "SH"."SALES")
ORDER BY DMR$CASE_ID