mixgb (version 1.5.2)

default_params_cran: Auxiliary function for validating xgb.params compatible with XGBoost CRAN version

Description

Auxiliary function for setting up the default XGBoost-related hyperparameters for mixgb and checking the xgb.params argument in mixgb(). For more details on XGBoost hyperparameters, please refer to XGBoost documentation on parameters.

Usage

default_params_cran(
  eta = 0.3,
  gamma = 0,
  max_depth = 3,
  min_child_weight = 1,
  max_delta_step,
  subsample = 0.7,
  sampling_method = "uniform",
  colsample_bytree = 1,
  colsample_bylevel = 1,
  colsample_bynode = 1,
  lambda = 1,
  alpha = 0,
  tree_method = "auto",
  max_leaves = 0,
  max_bin = 256,
  predictor = "auto",
  num_parallel_tree = 1,
  gpu_id = 0,
  nthread = -1
)

Value

A list of hyperparameters.

Arguments

eta

Step size shrinkage. Default: 0.3.

gamma

Minimum loss reduction required to make a further partition on a leaf node of the tree. Default: 0

max_depth

Maximum depth of a tree. Default: 3.

min_child_weight

Minimum sum of instance weight needed in a child. Default: 1.

max_delta_step

Maximum delta step. Default: 0.

subsample

Subsampling ratio of the data. Default: 0.7.

sampling_method

The method used to sample the data. Default: "uniform".

colsample_bytree

Subsampling ratio of columns when constructing each tree. Default: 1.

colsample_bylevel

Subsampling ratio of columns for each level. Default: 1.

colsample_bynode

Subsampling ratio of columns for each node. Default: 1.

lambda

L2 regularization term on weights. Default: 1.

alpha

L1 regularization term on weights. Default: 0.

tree_method

Options: "auto", "exact", "approx", and "hist". Default: "hist".

max_leaves

Maximum number of nodes to be added (Not used when tree_method = "exact"). Default: 0.

max_bin

Maximum number of discrete bins to bucket continuous features (Only used when tree_method is either "hist", "approx" or "gpu_hist"). Default: 256.

predictor

Default: "auto"

num_parallel_tree

The number of parallel trees used for boosted random forests. Default: 1.

gpu_id

Which GPU device should be used. Default: 0.

nthread

The number of CPU threads to be used. Default: -1 (all available threads).

Examples

default_params_cran()

xgb.params <- list(subsample = 0.9, gpu_id = 1)
default_params_cran(subsample = xgb.params$subsample, gpu_id = xgb.params$gpu_id)

xgb.params <- do.call("default_params_cran", xgb.params)
xgb.params