cre: Learn a compressed rule ensemble

In maltenlz/Compressed-Rule-Ensembles: R-Package to build Compressed Rule Ensembles

Learn a compressed rule ensemble

Description

Learn a compressed rule ensemble

Usage

cre(
  x,
  y,
  k = 4,
  treetype = "XGB",
  standardize = F,
  linear = T,
  alpha = 1,
  task = "regression",
  min_sup = 0.05,
  forest_control = NULL,
  model_type = "glmnet",
  eta = 0.5
)

Arguments

x	input matrix
y	outcome vector
treetype	foresttype to generate the rules. Currently XGBoost and RandomForest are supported
standardize	Should the rules be standardized? This essentiall removes the extra penalty on low support/high complexity rules. Can be sometimes beneficial.
linear	Boolean if linear terms should be included
alpha	alpha parameter in glmnet, alpha = 1 for lasso alpha = 0 for ridge.
task	"regression" or "class" for (binary) classification
forest_control	parameters that are passed to xgboost

Value

list containing the cre model

maltenlz/Compressed-Rule-Ensembles documentation built on Jan. 27, 2024, 6:43 a.m.