Are equivalent rules considered when testing in the Kover CV?

[augustkx]

Hi Alex,

Are equivalent rules considered when testing in the Kover CV? Seems not?

Best, Kaixin

[aldro61]

Hi Kaixin,

No, the equivalent rules are used purely for increased model interpretability (see example here). We do not use them in prediction. If you dig into the experimental code, you will see how we select rules in case of equivalence, but only one is retained for prediction and included in the final model.

Alex

[augustkx]

Ok, thanks, at least from the Fig 3, I see at least one reason why only one rule was selected among all.