grid.search.cross.validation: Grid search K-fold cross-validation

In Accelerytics/mlkit: Machine Learning Kit

Description

Implementation of the grid search approach using K-fold cross-validation for hyperparameter tuning of a given estimator.

Usage

grid.search.cross.validation(
  formula,
  data,
  estimator,
  params.list,
  n.folds = 5,
  ind.metric,
  comb.metric = mean,
  fold.id = NULL,
  force = F,
  verbose = F,
  plot = F,
  contour.scale = NULL,
  coef.lims = NULL,
  coef.names = NULL,
  seed = NULL,
  use.formula = T,
  ...
)

Arguments

formula	an object of class formula: a symbolic description of the model to be fitted following the standard of lm.
data	an optional data frame, list or environment (or object coercible by as.data.frame to a data frame) containing the variables in the model. If not found in data, the variables are taken from environment (formula), typically the environment from which this function is called.
estimator	estimator function that has arguments formula and data and returns a list containing parameters estimates in the component coefficients.
params.list	list or vector containing hyperparameters and their respective values to consider
n.folds	optional number of folds (K) in cross-validation. Default is 5.
ind.metric	metric function taking in a numerical vector of predictions for the dependent variable together with a numerical vector of actual outcomes of the dependent variable that returns a performance metric for the individual folds.
comb.metric	optional function used to combine individual fold metrics. Default is mean.
fold.id	optional vector containing numerical fold identifiers for each row in the data. If NULL, n.folds is used and random fold identifiers are constructed divided the observations equally over K folds. If provided, n.folds is ignored. Default is NULL.
force	optional boolean indicating whether or not to allow for errors due to singularity when applying the estimator. If TRUE, the individual metric of folds with hyperparameter combinations for which the estimator is not able to estimate coefficients due to singularity, are set to Inf and hence the errors are ignored. Default is FALSE.
verbose	optional boolean indicating whether to show a progress bar. Default is FALSE.
plot	optional boolean indicating whether to generate heatmaps of performance on the grid and displaying the estimated coefficients. Default is FALSE.
contour.scale	optional vector containing the hyperparameters and their scales for the axes in the contour plots. Default is NULL.
coef.lims	optimal limits of the coefficients plot. Default is NULL.
coef.names	optional names of coefficients, used in estimates barplot. Default is NULL in which case the column names of the explanatory variables are used.
seed	optimal seed to specify. Default is NULL.
use.formula	whether or not to use the formula data combination of X and y as inputs to the model. Default is TRUE.
...	additional arguments to be passed to the estimator function.

Value

grid.search.cross.validation returns an object of class "gscv". An object of class "gscv" is a list containing at least the following components:

coefficients	a named vector of optimal coefficients.
metric	metric of the optimal hyperparameters.
params	a named vector of optimal hyperparameters.

Accelerytics/mlkit documentation built on Dec. 31, 2020, 9:46 a.m.