-
Notifications
You must be signed in to change notification settings - Fork 3
Description
#51 (comment)
Hey! Recently, I ran into a similar issue as mentioned in the link while using SCM to select covariates. I had an nlmixr object named fit.1cpt.ode.focei. Then when I run
rxode2::.rxWithWd(tempdir(), {# with temporary directory
auto1 <- nlmixr2extra:::covarSearchAuto(fit.1cpt.ode.focei, varsVec = c("cl", "eta.cl"),
covarsVec = c("WT"))
}),
an error will occur and the error message is as follows:
── starting forward search...──────────────────────────────────────────────────────────────────────────
error while ADDING covariate
Error in backwardSearch(varsVec, covarsVec, catvarsVec, fitorig = fit, : cannot find the object 'fit'
After running fit <- fit.1cpt.ode.focei, calling nlmixr2extra:::covarSearchAuto again proceeded normally without any errors. However, when the fit object exists in the R workspace, even if I use fit.1cpt.ode.focei as the parameter when calling nlmixr2extra:::covarSearchAuto, the code still runs normally without errors. Once I remove fit from the workspace, the error "cannot find the object 'fit'" reappears.
Additionally, I've noticed that even after loading the nlmixr2extra package, I cannot directly use the covarSearchAuto() function—it does not appear in the help documentation, and only nlmixr2extra:::covarSearchAuto() runs without errors. Therefore, I would like to ask: is the SCM method currently ready for use, or is it still in a testing phase or something similar? Thanks!