engine {audrex}

support functions for audrex

Description

support functions for audrex

Usage

engine(
  predictors,
  target,
  booster,
  max_depth,
  eta,
  gamma,
  min_child_weight,
  subsample,
  colsample_bytree,
  lambda,
  alpha,
  n_windows,
  patience,
  nrounds
)

Arguments

predictors

A data frame with predictors on columns.

target

A numeric vector with target variable.

booster

String. Optimization methods available are: "gbtree", "gblinear". Default: "gbtree".

max_depth

Positive integer. Look to xgboost documentation for description. A vector with one or two positive integer for the search boundaries. The default value (NULL) sets automatically the values in c(1, 8).

eta

Positive numeric. Look to xgboost documentation for description. A vector with one or two positive numeric between (0, 1] for the search boundaries. The default value (NULL) sets automatically the values in c(0, 1).

gamma

Positive numeric. Look to xgboost documentation for description. A vector with one or two positive numeric for the search boundaries. The default value (NULL) sets automatically the values in c(0, 100).

min_child_weight

Positive numeric. Look to xgboost documentation for description. A vector with one or two positive numeric for the search boundaries. The default value (NULL) sets automatically the values in c(0, 100).

subsample

Positive numeric. Look to xgboost documentation for description. A vector with one or two positive numeric between (0, 1] for the search boundaries. The default value (NULL) sets automatically the values in c(0, 1).

colsample_bytree

Positive numeric. Look to xgboost documentation for description. A vector with one or two positive numeric between (0, 1] for the search boundaries. The default value (NULL) sets automatically the values in c(0, 1).

lambda

Positive numeric. Look to xgboost documentation for description. A vector with one or two positive numeric for the search boundaries. The default value (NULL) sets automatically the values in c(0, 100).

alpha

Positive numeric. Look to xgboost documentation for description. A vector with one or two positive numeric for the search boundaries. The default value (NULL) sets automatically the values in c(0, 100).

n_windows

Positive integer. Number of (expanding) windows for cross-validation. Default: 3.

patience

Positive numeric. Percentage of waiting rounds without improvement before xgboost stops. Default: 0.1

nrounds

Positive numeric. Number of round for the extreme boosting machine. Look to xgboost for description. Default: 100.

Author(s)

Giancarlo Vercellino [e-mail]