how to limit number of rules and to limit number of conditions in rules but get best performance for this constrains about imodels

[csinva]

Sorry for the super long delay and thanks for your interest 😄 ! We have been working to unify all the models into a solid base before dealing with details of individual models.

Bayesian Rule List is a very flexible model and this implementation only allows control over the parameters you listed above. In this case, the listwidth prior is the parameter that you most want to vary (the lower it is, the less rules). Limiting maxcardinality will help you get shorter rules. In either case, the algorithm will try to find the best list based on these hyperparameters!