next up previous contents
Next: Accuracy Estimation Up: Utilities Previous: Utilities


  The Inducer utility runs the given inducer on the given datafile and reports the following statistics:

Instance Counts
The number of training instances, test instances, the number of unseen test instances, and the number of instances seen.

Classification counts
The number of correct and incorrect classifications.

Generalization accuracy
The accuracy on the unseen instances.

Memorization accuracy
The accuracy on the seen instances. A big discrepancy between the generalization and memorization accuracy usually indicates overfitting.

The overall accuracy on the test set.

Figure 4: A snapshot of the MineSet tree visualizer fly-through.

Ronny Kohavi
Sun Oct 6 23:17:50 PDT 1996