Returns the subject-level covariates that `nlmixr2(..., est = "vae")` would
explore during automated covariate selection, using the same discovery rules
as the fit: every non-reserved numeric data column that is constant within
each subject is a candidate; a candidate with more than two unique values
(all positive) is treated as continuous (encoded `log(value/mean)`),
anything else as categorical (mean-centered). Time-varying numeric columns
cannot be searched and are excluded with a warning.
Usage
vaeCovariates(data, warn = TRUE)
Arguments
- data
estimation dataset containing at least an `ID` column; column
names are matched case-insensitively, as in the VAE fit
- warn
when `TRUE` (default) warn about time-varying numeric columns
excluded from the search; when `FALSE` exclude them silently
Value
a data frame with one row per explored covariate and columns
`covariate` (upper-cased column name), `type` (`"continuous"` or
`"categorical"`) and `center` (the population value the covariate is
centered at); zero rows when no covariates qualify
Examples
d <- data.frame(id = rep(1:3, each = 2), time = rep(0:1, 3), dv = rnorm(6),
wt = rep(c(70, 80, 60), each = 2),
sex = rep(c(0, 1, 0), each = 2))
vaeCovariates(d)
#> covariate type center
#> 1 WT continuous 70.0000000
#> 2 SEX categorical 0.3333333
