Providing software solutions since 1976

• Log in
• Create profile
• Welcome
  [Log out]
• Edit profile

• Print  |
• Feedback  |
• Share

  • E-mail

  
  
  

  E-mail This Page

  Your Name:*
  Your E-mail:*
  Recipient E-mail:*
  Send a copy to me
  Personal Message:
  
  

  * indicates a required field

Predicted probabilities output from the Decisions node of SAS^® Enterprise Miner^TM might be incorrect in any of these three situations:

1. A flow with an Input Data Source node followed by a Regression node and a Decisions node will return incorrect scorecode and prior probabilities when the Input Data Source has priors defined, and the Decisions node has new priors defined.

   The score code will appear similar to this example (good_bad is the target variable with levels good and bad):

      *------------------------------------------------------------*;
      *Posterior Probabilities: good_bad;
      *------------------------------------------------------------*;
        1 = P_good_badgood * 0.7 / 0.9;
        2 = P_good_badbad * 0.3 / 0.1;
   

   However, it should say:

      *------------------------------------------------------------*;
      *Posterior Probabilities: good_bad;
      *------------------------------------------------------------*;
        P_good_badgood = P_good_badgood * 0.7 / 0.9;
        P_good_badbad = P_good_badbad * 0.3 / 0.1;
   

2. In a similar flow, the updating posterior probabilities section of the score code is incorrect. The posterior probabilities are not correctly adjusted when modifying existing priors in the Decisions node (that is, prior probabilities were used when modeling the target, and new ones are specified in the Decision nodes.) This affects the posterior probabilities created by the Decisions node, and thus the assessment results of the new model will not be correct.
3. A baseline model is created by adding a Decisions node without a preceding modeling node. If you have previously defined priors and bring up the Decisions window in the Decisions node without modifying the prior probabilities, then the posterior probabilities for the baseline model will not be adjusted with the prior probabilities.

Select the Hot Fix tab in this note to access the hot fix for this issue.

Operating System and Release Information

A fix for this issue for SAS Enterprise Miner Server 5.2 is available at: