AutoCompBoostBase: AutoCompBoostBase

In Coorsaa/autocompboost: AutoML with compboost

AutoCompBoostBase

Description

Base class for AutoCompBoost. Has subclasses for Classification and Regression.

Creates a new instance of this R6 class.

Trains the AutoML system.

Returns a Prediction object for the given data based on the trained model.

Performs nested resampling if enable_tuning equals TRUE.

Helper to extract the best hyperparameters from a tuned model.

Returns the trained model if final_model is set to TRUE.

Returns the resample result of method resample().

Returns the risk stages of the final model.

Plot function to plot the final model's risk stages.

Plot function to plot the feature importance.

Plot function to plot the feature importance.

Plot function to plot the univariate effects.

Plot function to plot the interaction effects.

Returns the selected base learners by the final model.

Plot function to plot a single spline.

Plot function to plot the learner traces.

Arguments

task	(Task) Contains the task to be solved. Currently TaskClassif and TaskRegr are supported.
resampling	(Resampling) Contains the resampling method to be used for hyper-parameter optimization. Defaults to ResamplingCV with 3 folds.
param_values	(list()) Parameter values which are pass on to the learner.
measure	(Measure) Contains the performance measure, for which we optimize during training. Defaults to Accuracy for classification and RMSE for regression.
tuning_method	(character(1)) Tuning method. Possible choices are "mbo", "hyperband" or "smash"¸ Default is "mbo".
tuning_time	(integer(1)) Termination criterium. Number of seconds for which to run the optimization. Does not include training time of the final model. Default is set to 3600, i.e. one hour. Tuning is terminated depending on the first termination criteria fulfilled.
tuning_iters	(integer(1)) Termination criterium. Number of MBO iterations for which to run the optimization. Default is set to 150 iterations. Tuning is terminated depending on the first termination criteria fulfilled.
tuning_generations	(integer(1)) Termination criterium for tuning method smashy. Number of generations for which to run the optimization. Default is set to 3 generations. Tuning is terminated depending on the first termination criteria fulfilled.
enable_tuning	(logical(1)) Whether or not to perform hyperparameter optimization. Default is TRUE.
final_model	(logical(1)) Whether or not to return the final model trained on the whole dataset at the end.
data	(data.frame | data.table | Task) New observations to be predicted. If NULL, defaults to the task the model was trained on.
row_ids	(integer()) Vector of training indices.
outer_resampling	(Resampling) Resampling strategy. Default is ResamplingHoldout for the outer resampling.
spline	(character(1L)) Name of spline to plot.

Value

AutoCompBoostBase

PredictionClassif | PredictionRegr

ResampleResult

data.table

Compboost

mlr3

(list)

patchwork

patchwork

patchwork

patchwork

patchwork

(character())

ggplot

ggplot

Fields

task

(Task)
Contains the task to be solved.

learner

(AutoTuner)
The ML pipeline at the core of mlr3automl is an AutoTuner containing a GraphLearner.

param_values

(list())
Parameter values which are pass on to the learner.

resampling

(Resampling)
Contains the resampling method to be used for hyper-parameter optimization.

measure

(Measure)
Contains the performance measure, for which we optimize during training.

tuning_method

(character(1))
Tuning method. Possible choices are "mbo", "hyperband" or "smashy"¸ Default is "smashy".

tuning_time

(integer(1))
Termination criterium. Number of seconds for which to run the optimization. Does not include training time of the final model.
Default is set to 60, i.e. one minuet. Tuning is terminated depending on the first termination criteria fulfilled.

tuning_iters

(integer(1))
Termination criterium. Number of MBO iterations for which to run the optimization.
Default is set to 150 iterations. Tuning is terminated depending on the first termination criteria fulfilled.

tuning_generations

(integer(1))
Termination criterium for tuning method smashy. Number of generations for which to run the optimization.
Default is set to 3 generations. Tuning is terminated depending on the first termination criteria fulfilled.

enable_tuning

(logical(1))
Whether or not to perform hyperparameter optimization. Default is TRUE.

final_model

(logical(1))
Whether or not to return the final model trained on the whole dataset at the end.

tuner

(TunerInterMBO)
Tuning is performed using TunerInterMBO.

tuning_terminator

(Terminator)
Contains an termination criterion for model tuning.

Internals

The AutoCompBoostBase class uses mlr3pipelines to create a machine learning pipeline.
This pipeline contains multiple preprocessing steps wrapped in a GraphLearner.
This GraphLearner is wrapped in an AutoTuner for Hyperparameter Optimization and proper resampling.
Tuning is performed using Bayesian Optimization.

Construction

Objects should be created using the AutoCompBoost interface function.

model = AutoCompBoost(task, resampling, measure, tuning_budget, tuning_iters, final_model)

Coorsaa/autocompboost documentation built on March 19, 2023, 12:08 p.m.