The Model Comparison node Results window in SAS^® Enterprise Miner™ displays an Average Squared Error statistic in the Fit Statistics panel. If your data is partitioned into train, validate, and test, then the Model Comparison node displays an Average Squared Error statistic for each partition. The labels should be displayed as follows:

• Train: Average Squared Error
• Valid: Average Squared Error
• Test: Average Squared Error

However, the "Test: Average Squared Error" label is incorrectly displayed as "Train: Average Squared Error". As a result, two columns labeled "Train: Average Squared Error" are displayed. The problem occurs when a Decision Tree node is included in the Model Comparison results. There are no errors or warnings to indicate that the label is not correct.

The "Train: Average Squared Error" column that is correctly labeled appears with the other Train result columns. The "Train: Average Squared Error" column that is incorrectly labeled appears with the Test result columns. Here is an example of the incorrectly labeled column.

The Average Squared Error values are not incorrect. Only the label is incorrect.

The Average Squared Error statistics are correctly labeled in the Output panel.

Operating System and Release Information