LearnerLightgbm {mllrnrs}

R6 Class to construct a LightGBM learner

Description

The LearnerLightgbm class is the interface to the lightgbm R package for use with the mlexperiments package.

Details

Optimization metric: needs to be specified with the learner parameter metric. The following options can be set via options():

• "mlexperiments.optim.lgb.nrounds" (default: 5000L)

• "mlexperiments.optim.lgb.early_stopping_rounds" (default: 500L)

• "mlexperiments.lgb.print_every_n" (default: 50L)

• "mlexperiments.lgb.verbose" (default: -1L)

LearnerLightgbm can be used with

• mlexperiments::MLTuneParameters

• mlexperiments::MLCrossValidation

• mlexperiments::MLNestedCV

Super class

mlexperiments::MLLearnerBase -> LearnerLightgbm

Methods

Public methods

• LearnerLightgbm$new()

• LearnerLightgbm$clone()

Inherited methods

• mlexperiments::MLLearnerBase$bayesian_scoring_function()
• mlexperiments::MLLearnerBase$cross_validation()
• mlexperiments::MLLearnerBase$fit()
• mlexperiments::MLLearnerBase$predict()

Method new()

Create a new LearnerLightgbm object.

Usage

LearnerLightgbm$new(metric_optimization_higher_better)

Arguments

metric_optimization_higher_better

A logical. Defines the direction of the optimization metric used throughout the hyperparameter optimization.

Returns

A new LearnerLightgbm R6 object.

Examples

LearnerLightgbm$new(metric_optimization_higher_better = FALSE)

Method clone()

The objects of this class are cloneable with this method.

Usage

LearnerLightgbm$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

See Also

lightgbm::lgb.train(), lightgbm::lgb.cv()

Examples

# binary classification

library(mlbench)
data("PimaIndiansDiabetes2")
dataset <- PimaIndiansDiabetes2 |>
  data.table::as.data.table() |>
  na.omit()

seed <- 123
feature_cols <- colnames(dataset)[1:8]

param_list_lightgbm <- expand.grid(
  bagging_fraction = seq(0.6, 1, .2),
  feature_fraction = seq(0.6, 1, .2),
  min_data_in_leaf = seq(10, 50, 10),
  learning_rate = seq(0.1, 0.2, 0.1),
  num_leaves = seq(10, 50, 10),
  max_depth = -1L
)

train_x <- model.matrix(
  ~ -1 + .,
  dataset[, .SD, .SDcols = feature_cols]
)
train_y <- as.integer(dataset[, get("diabetes")]) - 1L

fold_list <- splitTools::create_folds(
  y = train_y,
  k = 3,
  type = "stratified",
  seed = seed
)
lightgbm_cv <- mlexperiments::MLCrossValidation$new(
  learner = mllrnrs::LearnerLightgbm$new(
    metric_optimization_higher_better = FALSE
  ),
  fold_list = fold_list,
  ncores = 2,
  seed = 123
)
lightgbm_cv$learner_args <- c(
  as.list(
    data.table::data.table(
      param_list_lightgbm[37, ],
      stringsAsFactors = FALSE
    ),
  ),
  list(
    objective = "binary",
    metric = "binary_logloss"
  ),
  nrounds = 45L
)
lightgbm_cv$performance_metric_args <- list(positive = "1")
lightgbm_cv$performance_metric <- mlexperiments::metric("auc")

# set data
lightgbm_cv$set_data(
  x = train_x,
  y = train_y
)

lightgbm_cv$execute()


## ------------------------------------------------
## Method `LearnerLightgbm$new`
## ------------------------------------------------

LearnerLightgbm$new(metric_optimization_higher_better = FALSE)