Skip to contents

Train Supervised Learning Models

Source: R/train.R
train.Rd

Preprocess, tune, train, and test supervised learning models with a single function using nested resampling

Usage

train(
  x,
  dat_validation = NULL,
  dat_test = NULL,
  algorithm = NULL,
  preprocessor_config = NULL,
  hyperparameters = NULL,
  tuner_config = NULL,
  outer_resampling_config = NULL,
  weights = NULL,
  question = NULL,
  outdir = NULL,
  parallel_type = c("future", "mirai", "none"),
  future_plan = getOption("future.plan", "multicore"),
  n_workers = max(future::availableCores() - 3L, 1L),
  verbosity = 1L
)

Arguments

x

data.frame or similar: Training set data.

dat_validation

data.frame or similar: Validation set data.

dat_test

data.frame or similar: Test set data.

algorithm

Character: Algorithm to use. Can be left NULL, if hyperparameters is defined.

preprocessor_config

PreprocessorConfig object or NULL: Setup using setup_Preprocessor.

hyperparameters

Hyperparameters object: Setup using one of setup_* functions.

tuner_config

TunerConfig object: Setup using setup_GridSearch.

outer_resampling_config

ResamplerConfig object or NULL: Setup using setup_Resampler. This defines the outer resampling method, i.e. the splitting into training and test sets for the purpose of assessing model performance. If NULL, no outer resampling is performed, in which case you might want to use a dat_test dataset to assess model performance on a single test set.

weights

Optional vector of case weights.

question

Optional character string defining the question that the model is trying to answer.

outdir

Character, optional: String defining the output directory.

parallel_type

Character: "none", "future", or "mirai".

future_plan

Character: Future plan to use for parallel processing.

n_workers

Integer: Number of workers to use for parallel processing in total. Parallelization may happen at three different levels, from innermost to outermost:

  1. Algorithm training (e.g. a parallelized learner like LightGBM)

  2. Tuning (inner resampling, where multiple resamples can be processed in parallel)

  3. Outer resampling (where multiple outer resamples can be processed in parallel) The train() function will assign the number of workers to the innermost available parallelization level. Best to leave a few cores for the OS and other processes, especially on shared systems or when working with large datasets, since parallelization will increase memory usage.

verbosity

Integer: Verbosity level. hyperparameters is not defined. Avoid relying on this, instead use the appropriate setup_* function with the hyperparameters argument.

Value

Object of class Regression(Supervised), RegressionRes(SupervisedRes), Classification(Supervised), or ClassificationRes(SupervisedRes).

Details

Important: For binary classification, the outcome should be a factor where the 2nd level corresponds to the positive class.

Note on resampling: You should never use an outer resampling method with replacement if you will also be using an inner resampling (for tuning). The duplicated cases from the outer resampling may appear both in the training and test sets of the inner resamples, leading to underestimated test error.

Author

EDG