Inner engine shared by all the cross-validation drivers: splits the data into folds, loops over a hyper-parameter grid fitting a model per (grid value, fold) pair via fit_fun, reconstructs the test predictions via reconstruct_fun, and collects the quantile error matrix.
Usage
cv_framework(
Y,
grid,
n.folds,
criteria,
quantile.value,
fit_fun,
reconstruct_fun,
return.models = TRUE,
verbose.cv = TRUE,
seed = NULL,
grid.label = "Degrees of freedom: ",
model.prefix = "df_idx"
)Arguments
- Y
The \((N \times T)\) matrix of functional data.
- grid
Numeric vector of hyper-parameter values to cross-validate.
- n.folds
Number of folds to be used on cross validation.
- criteria
Criteria used to divide the data. Valid values are
'rows'or'points'.- quantile.value
The quantile considered.
- fit_fun
Function
(Y.train, grid.value, train.idx)returning a fitted model.train.idxcontains the training row indices whencriteria='rows'and is NULL otherwise.- reconstruct_fun
Function
(model, Y.test, train.idx)returning the predicted \((N \times T)\) matrix used to score the fold.- return.models
Should the list of all the models built be returned?
- verbose.cv
Boolean indicating verbosity of the cross-validation process.
- seed
Seed for the random generator number (used to build the folds).
- grid.label
Label prefixing the grid value in verbose messages.
- model.prefix
Prefix used to name the stored models (
'<prefix>=<i>.fold=<j>').