How many levels is a class variable allowed to have before being marked as REJECTED by default?

A class variable in SAS Enterprise Miner is a categorical variable that can represent different levels or categories. By default, if a class variable has more levels than a certain cutoff, it may be marked as rejected because it could complicate the modeling process and lead to issues like overfitting or complexity in the interpretation of results.

In the context of SAS Enterprise Miner, the cutoff for the maximum number of levels that a class variable can have before being rejected is set at 20. This limit helps to ensure that the model remains interpretable and manageable without introducing excessive noise from too many levels. Having a large number of categories might dilute the strength of the relationships being modeled, hence the necessity for such a threshold.

In practical applications, if the number of unique levels exceeds this default limit, users would need to either combine some categories, reduce the number of categories, or employ different strategies to handle high cardinality in categorical variables. Understanding this limit helps with producing more efficient and effective models in SAS Enterprise Miner.