ez_benchmark: Tunes a model via 'EZtune' and returns the benchmarking...

In jillbo1000/EZtuneTest:

Description

ez_benchmark tunes a model with eztune and returns a one row data.frame with the results. rsample is used to split the dataset into training and test datasets. The results returned

Usage

ez_benchmark(
  x,
  y,
  name = "Data",
  method = "svm",
  optimizer = "hjn",
  fast = TRUE,
  cross = NULL,
  loss = "default"
)

Arguments

x	Matrix or data frame containing the dependent variables.
y	Vector of responses. Can either be a factor or a numeric vector.
name	Name of the dataset. Used to name the output file and as an identifier within the output dataset.
method	Model to be fit. Choices are "ada" for adaboost, "en" for elastic net, "gbm" for gradient boosting machines, and "svm" for support vector machines. This argument is passed to eztune. See EZtune::eztune for more information.
optimizer	Optimization method to be used. May be either "hjn" or "ga". This argument is passed to eztune. See EZtune::eztune for more information.
fast	Value for fast argument in eztune. Options are TRUE, FALSE, a positive integer, or a number between 0 and 1. See EZtune::eztune for more information.
cross	Value for cross argument in eztune.Options are NULL or a positive integer to represent the number of folds. See EZtune::eztune for more information.
loss	Value for loss argument in eztune. Options are "default", "class", "auc", "mse", or "mae". See EZtune::eztune for more information.

Value

Saves a matrix to the indicated path that contains the results for each of the runs. The final file contains the following variables:

data	Name of the dataset.
package	This will always be "EZtune".
method	Type of model that was fit. It will "gbm" for gradient boosting machines, "svm" for support vector machines, "en" for elastic net, or "ada" for adaboost.
optimizer	Optimization method. It will be "Genetic algorithm" or "Hooke-Jeeves".
assess	The method used to assess the model during tuning. It will be either cross-validation or some version of FAST.
tuned_on	Metric that was used as the loss function.
acc_rmse	The best accuracy or RMSE obtained from the model as determined by a validation dataset.
auc_mae	The best AUC or MAE obtained from the model as determined by a validation dataset.
time	Time to complete the calculations in seconds.

The models are verified using rsample to split the data into training and testing datasets.

See Also

tm_benchmark, EZtune::eztune

jillbo1000/EZtuneTest documentation built on Oct. 5, 2021, 4:16 p.m.