Data mining problems routinely involve situations where one target level is more "rare" than others. By default, SAS Enterprise Miner assigns the most likely outcome as the predicted outcome. This assignment results in decision rules that strongly favor the common outcome, which is usually not of interest. The assignment often generates models with no predicted events of interest.

If you specify priors, then the posterior probabilities are adjusted, but the adjustment might lead to no variables selected. Even when a model is successfully fit, the predicted outcome might be the common target level.

Example: An event that occurs 1% of the time.  A person who is 10 times as likely as average to have the event still has only a 10% chance of having the event.  But, he has a 90% chance of having the non-event.

You can change this prediction outcome by modifying the default decision-weights. Edit the default decision-weights either in the Input Data node (or in a Decisions node prior to using a modeling node).

To modify by specifying Inverse Prior Weights in the Input Data node, follow these steps:

1. Click the Input Data node.
2. Click the ellipsis (...) for the Decisions property.
3. Click Build to create a target profile.
4. Click the Decisions tab.

5. Click Default with Inverse Prior Weights. This selection enables you to find variables that are useful predictors.

6. Click Decision Weights to see that the values are changed from their default values.
7. Click OK.

For more information, see the chapter "Predictive Modeling" in SAS Enterprise Miner Help.

Operating System and Release Information