Hi! 
I am new to Oracle Data Mining. I hope anyone here is also using ODM.
I created 10 model simulations (same building steps). Since ODM shuffles data before splitting it into build and test cases, I got different test metrics (with difference not so significant though). I would like to know what method I should look at to choose the best model.
There were models with high predictive confidence, but their (Accuracy/ROC) overall accuracy is low (compared to other models). So, I'm kinda confused which one to choose. :| Would predictive confidence be just enough as an indicator of a good model?
Thank you!
I am new to Oracle Data Mining. I hope anyone here is also using ODM.
I created 10 model simulations (same building steps). Since ODM shuffles data before splitting it into build and test cases, I got different test metrics (with difference not so significant though). I would like to know what method I should look at to choose the best model.
There were models with high predictive confidence, but their (Accuracy/ROC) overall accuracy is low (compared to other models). So, I'm kinda confused which one to choose. :| Would predictive confidence be just enough as an indicator of a good model?
Thank you!