I am using oracle data miner 11g for building association model (market basket analysis). When I run the model and view the results, I find that the values of Antecedent Support% and Consequent Support% are swapped. i.e. the support value of the antecedent is listed in the Consequent Support% column.
Is it a bug in the data miner or am I mistaken?
Edited by: 976043 on Dec 10, 2012 10:00 AM
Yes you are correct.
You can prove it my adding a Model Details node and connecting the AR Build node to it.
The rule output from the Model Details node produced shows the true values.
We have opened a bug to fix this and it will be in the upcoming SQL Dev 4.0 release.
Thanks for the help, Mark