predict_SL.PredictionStack: Predict for discrete SuperLearner fit
In osofr/gridisl: Discrete Super Learner with Grid-Search for Longitudinal Data

Predict for discrete SuperLearner fit

## S3 method for class 'PredictionStack'
predict_SL(modelfit, newdata,
  add_subject_data = FALSE, subset_idx = NULL, holdout = FALSE,
  verbose = getOption("gridisl.verbose"), ...)

`modelfit`	Model fit object returned by `fit`.
`newdata`	Subject-specific data for which predictions should be obtained. If missing then the predictions for the training data will be typically returned. See `holdout` for discussion of alternative cases.
`add_subject_data`	Set to `TRUE` to add the subject-level data to the resulting predictions (returned as a data.table). When `FALSE` (default) only the actual predictions are returned (as a matrix with each column representing predictions from a specific model).
`subset_idx`	A vector of row indices in `newdata` for which the predictions should be obtain. Default is `NULL` in which case all observations in `newdata` will be used for prediction.
`holdout`	Set to `TRUE` for out-of-sample predictions for validation folds (out-of-sample observations) or holdouts. When `newdata` is missing there are two possible types of holdout predictions, depending on the modeling approach. 1. For `method = "holdout"` the default holdout predictions will be based on validation data. 2. For `method = "cv"` the default is to leave use the previous out-of-sample (holdout) predictions from the training data.
`verbose`	Set to `TRUE` to print messages on status and information to the console. Turn this on by default using `options(gridisl.verbose=TRUE)`.
`...`	Additional arguments, used only for compatibility with other `predict_SL` functions.

A data.table of subject level predictions (subject are rows, columns are different models) or a data.table with subject level covariates added along with model-based predictions.

osofr/gridisl documentation built on May 24, 2019, 4:55 p.m.