LightRuleFitHyperparameters

Properties

nroundstunable

integerdefault 200≥ 1

Number of boosting rounds (LightGBM step).

num_leavestunable

integerdefault 32≥ 1

Maximum number of leaves in one tree (LightGBM step).

max_depthtunable

integerdefault 4

Maximum tree depth (LightGBM step). -1 = no limit.

learning_ratetunable

numberdefault 0.1≤ 1> 0

Learning rate (LightGBM step).

subsampletunable

numberdefault 0.666≤ 1> 0

Fraction of cases sampled per tree (LightGBM step).

subsample_freqtunable

integerdefault 1≥ 1

Bagging frequency (LightGBM step).

lambda_l1tunable

numberdefault 0≥ 0

L1 regularization (LightGBM step).

lambda_l2tunable

numberdefault 0≥ 0

L2 regularization (LightGBM step).

objective

string | nulldefault null

LightGBM objective. NULL = set from outcome type.

ifw_lightgbmtunable

booleandefault false

Inverse Frequency Weighting in the LightGBM step.

alphatunable

numberdefault 1≥ 0≤ 1

Elastic net mixing parameter (GLMNET step).

lambda

array | nulldefault null≥ 1 items

Regularization strength (GLMNET step). NULL = determined by cv.glmnet.

ifw_glmnettunable

booleandefault false

Inverse Frequency Weighting in the GLMNET step.

ifw

booleandefault false

Inverse Frequency Weighting in both steps. Cannot be combined with ifw_lightgbm or ifw_glmnet.

Relationships