Re: Applied Analytics Using SAS Enterprise Miner
At page 6-25 of the course notes, the note at the bottom points out the need to amend the Model Selection Criterion when "Decision Data" is defined.
However, up to that point, only prior probabilities had been entered (page 6-24): my understanding is that "Decision Data" refers to specifying Decision Weights; prior probabilities on their own would not trigger Enterprise Miner to switch using Profit/Loss as an assessment measure, as page 6-25 seems to imply.
Or is my understanding incorrect?
My Answer:
Yes your understanding is correct. Adjusting for Prior will only adjust the posterior probability. But the decision threshold is still at 0.5. To change the cut point probability corresponding to the priors, decision weights need to be included and use decision should be checked in the third Decision Tab.