|
Setting
|
Explanation
|
|
Estimation method for the
prediction accuracy
|
| Requires additionally the specification of the data splitting used to divide the learn data sample into tuning learn and test data samples.
|
| · | Leave-one-out cross-validation (LOOCV)
|
|
|
Skipping
|
Skip parameter sets, if the resulting model size will exceed the maximum size or the minimum compression thresholds
|
|
Maximum size
|
Maximum model size relative to the number of tuning learn data tuples
|
|
Min. compression
|
Maximum model size relative to the number of tuning learn data tuples without elimination of rules with a small mass
|
|
Randomize
|
Initialize the random number generator.
Hint: Sometimes it is quite useful to turn this off, if you want to have reproducible results.
|