https://communities.sas.com/t5/SAS-Data-Science/AUC-in-Enterprise-Miner-different-in-Scorecard-and-Model/m-p/462064#M7002 <P>Update:&nbsp;</P><P>I was using the KGB target (i.e., allowing for missing target values). The scorecard node produced the AUC based on the KGB target. The model comparison node provided the AUC based on recoding the target to be AGB (i.e., setting the missing target values equal to zero).&nbsp;</P><P>&nbsp;</P><P>I found this out by running my own c-stat macro in Enterprise Guide using the KGB and AGB targets.</P> Mon, 14 May 2018 14:40:06 GMT acarman1 2018-05-14T14:40:06Z https://communities.sas.com/t5/SAS-Data-Science/AUC-in-Enterprise-Miner-different-in-Scorecard-and-Model/m-p/461607#M6990 <P>I'm building a scorecard in Enterprise Miner 14.2. If I right-click on the scorecard node and view the results, I see an AUC of 0.73. However, when I run the Model Comparison node, the scorecard has an AUC of 0.61. All paths are updated and I've re-run the diagram from start to finish. I don't suspect that a difference in computation methods could lead to such a gap in AUCs. Any idea what's going on?</P> Fri, 11 May 2018 16:38:59 GMT https://communities.sas.com/t5/SAS-Data-Science/AUC-in-Enterprise-Miner-different-in-Scorecard-and-Model/m-p/461607#M6990 acarman1 2018-05-11T16:38:59Z https://communities.sas.com/t5/SAS-Data-Science/AUC-in-Enterprise-Miner-different-in-Scorecard-and-Model/m-p/462064#M7002 <P>Update:&nbsp;</P><P>I was using the KGB target (i.e., allowing for missing target values). The scorecard node produced the AUC based on the KGB target. The model comparison node provided the AUC based on recoding the target to be AGB (i.e., setting the missing target values equal to zero).&nbsp;</P><P>&nbsp;</P><P>I found this out by running my own c-stat macro in Enterprise Guide using the KGB and AGB targets.</P> Mon, 14 May 2018 14:40:06 GMT https://communities.sas.com/t5/SAS-Data-Science/AUC-in-Enterprise-Miner-different-in-Scorecard-and-Model/m-p/462064#M7002 acarman1 2018-05-14T14:40:06Z