fit.ModelStack: Fit Discrete SuperLearner
In osofr/longGriDiSL: Discrete Super Learner with Grid-Search for Longitudinal Data

Description Usage Arguments Value

Define and fit discrete SuperLearner for longitudinal data. Model selection (scoring) can be based on MSE evaluated for random holdout observations (method = "holdout") or V-fold cross-validated MSE (method = "cv").

fit(...)

## S3 method for class 'ModelStack'
fit(models, method = c("none", "holdout", "cv",
  "origamiSL", "internalSL"), data, ID, t_name, x, y, nfolds = NULL,
  fold_column = NULL, hold_column = NULL, hold_random = FALSE,
  seed = NULL, refit = TRUE, fold_y_names = NULL,
  verbose = getOption("gridisl.verbose"), ...)

`...`	Additional arguments that will be passed on directly to `fit_model` function.
`models`	Parameters specifying the model(s) to fit. This must be a result of calling `defModel(...) + defModel(...)` functions. See `defModel` for additional information.
`method`	The type of model selection and model stacking procedure when fitting more than one model. Possible options are: `"none"` – no model selection; `"holdout"` – model selection based on a (possibly) random holdout validation sample; `"cv"` – discrete Super Learner, select a single best-performing model via internal V-fold cross-validation; `"origamiSL"` – convex (NNLS) Super Learner with external V-fold cross-validation (using `origami` R package); `"internalSL"` – convex (NNLS) Super Learner with internal V-fold cross-validation (same CV as in method="cv");
`data`	Input dataset, can be a `data.frame` or a `data.table`.
`ID`	A character string name of the column that contains the unique subject identifiers.
`t_name`	A character string name of the column with integer-valued measurement time-points (in days, weeks, months, etc).
`x`	A vector containing the names of predictor variables to use for modeling. If x is missing, then all columns except `ID`, `y` are used.
`y`	A character string name of the column that represent the response variable in the model.
`nfolds`	Number of folds to use in cross-validation.
`fold_column`	The name of the column in the input data that contains the cross-validation fold indicators (must be an ordered factor).
`hold_column`	The name of the column that contains the holdout observation indicators (TRUE/FALSE) in the input data. This holdout column must be defined and added to the input data prior to calling this function.
`hold_random`	Logical, specifying if the holdout observations should be selected at random. If FALSE then the last observation for each subject is selected as a holdout.
`seed`	Random number seed for selecting random holdouts or validation folds.
`refit`	Set to `TRUE` (default) to refit the best estimator using the entire dataset. When `FALSE`, it might be impossible to make predictions from this model fit.
`fold_y_names`	(ADVANCED FEATURE) The names of columns in `data` containing the fold-specific outcomes. Can be used for contructing split-specific (or by-fold) Super-Learner with `method`=`"origamiSL"`.
`verbose`	Set to `TRUE` to print messages on status and information to the console. Turn this on by default using `options(gridisl.verbose=TRUE)`.