add_predictions() with categorical variables (factors) in nls model

[rungec]

Hi, When adding nls model predictions to tibble, where nls model includes categorical variable (a factor) add_predictions() seems to assume the tibble is ordered by that factor. see example below. Works fine for glm models though, so wasn't sure whether to post here or on the nls pages. Claire

require(tidyverse)

mtcars_tibble <- as_tibble(mtcars) %>% mutate(cyl=as.factor(cyl))

lf <- formula(mpg ~ exp(a+b*wt)| cyl)
mod1 <- nlme::nlsList(lf, data=mtcars_tibble, start=list(a=0,b=0))

mtcars_preds <- mtcars_tibble %>% add_predictions(mod1)

ggplot(mtcars_preds, aes(y=mpg, x=wt, group=cyl))+
  geom_point() +
  geom_line(aes(y=pred), col="red") +
  facet_wrap("cyl")

#adding arrange fixes the problem
mtcars_preds <- mtcars_tibble %>% arrange(cyl) %>% add_predictions(mod1)

ggplot(mtcars_preds, aes(y=mpg, x=wt, group=cyl))+
  geom_point() +
  geom_line(aes(y=pred), col="red") +
  facet_wrap("cyl")

[hadley]

modelr is now superseded, which means that we'll only perform critical bug fixes needed to keep it on CRAN. Thanks for contributing this idea and my apologies that it took so long to inform you that this package is no longer under development.