MBR prediction error and confusional matrix

[lele241]

How to find the total predictional error on the test set and the confusional matrix on validation set for MBR classificator on SAS enterprise miner?

  Answere asap.

             thank you very much

[fef92]

I have the same problem :smileyshocked:

[gergely_batho]

Connect a Model Comparison Node to MBR.

There you should see it.

[fef92]

I found the confusional matrix, but not the error of test set

[gergely_batho]

In Model Comparison results look at the Fit Statistics table.

Some of the last columns are: Test ASE, Test RASE, Test MSE, Test RMSE, Test Misc.Rate, etc.

Make also sure, that you have a Partition Node in the flow, and test set percent is not 0%.

[fef92]

Thank you very much!

In the Partition Node test set percent is not 0%, but the value of final prediction error is " . " .I had a filter node for the target before the Partition, maybe I have deleted something important...

[gergely_batho]

As a first step try to connect a Model Comparison Node directly to the MBR node.

What does your Filter Node do? Hopefully not dropping all the test observation.

You can select the Model Comparison Node, then click on Imported Data, and Browse directly the Test Data. Does it have predictions (not missing)? Is the target variable OK (not missing)?

Are you saying that Test ASE, etc are all missing in the Fit Statistic Table?