Calculate fold errors for a fitted `pk` object.
# S3 method for pk
fold_error(
obj,
newdata = NULL,
model = NULL,
method = NULL,
exclude = TRUE,
sub_pLOQ = TRUE,
suppress.messages = NULL,
...
)A data.frame with one row for each `data_group`, `model` and `method`. A column contains the fold errors (observed/predicted) of the model fitted by the corresponding method. These residuals are concentrations in the same units as `obj$data$Conc.Units`; any concentration transformations (in `obj$scale$conc`) are *not* applied.
A `pk` object
Optional: A `data.frame` with new data for which to compute fold errors. If NULL (the default), then fold errors will be computed for the data in `obj$data`. `newdata` is required to contain at least the following variables: `Time`, `Dose`, `Route`, and `Media`.
Optional: Specify one or more of the fitted models for which to make predictions and calculate fold errors. If NULL (the default), fold errors will be returned for all of the models in `obj$stat_model`.
Optional: Specify one or more of the [optimx::optimx()] methods for which to make predictions and calculate RMSEs. If NULL (the default), fold errors will be returned for all of the models in `obj$settings_optimx$method`.
Logical: `TRUE` to return `NA_real_` for any observations in the data marked for exclusion (if there is a variable `exclude` in the data, an observation is marked for exclusion when `exclude `FALSE` to return the prediction for each observation, regardless of exclusion. Default `TRUE`.
Logical: whether or not to include predictions below pLOQ. when TRUE, values below pLOQ will be replaced by pLOQ.
Logical: whether to suppress message printing. If NULL (default), uses the setting in `object$settings_preprocess$suppress.messages`
Additional arguments. Currently not in use.
Caroline Ring
Here, fold error is defined as `observed/predicted`.
# Scaling and transformation of concentration variables in `newdata`
This function differs from some of the other methods for a fitted [pk()] object that accept `newdata`, in that there is no `use_scale_conc` argument for [fold_error.pk()]. Fold errors are always computed on the natural, un-transformed concentration scale (but note that fold error on a dose-normalized scale will be the same as fold error on a non-dose-normalized scale).