Use map to loop over multiple variables to regress in nested dataset - r

Question

I have a nested dataset, where each row contains in a column a tibble with a subset of the main dataset.
I want to run regressions over each of these subsets, where I vary the dependent variable.
I built a function to do this taking only one variable as the dependent variable.
I'm trying to use map to loop over multiple possible dependent variables but it seems as if the formula is not being interpreted correctly.
library(purrr)
library(tidyr)
library(rlang)
reg_across_groups <- function(df, var) {
df <- df %>%
nest(data = -c("cyl"))
var_enq <- rlang::enquo(var)
model_formula <- formula(paste0(rlang::quo_name(var_enq), "~ hp"))
df %>%
dplyr::mutate(model = purrr::map(data, ~lm(model_formula, data = .x)))
}
# Works
reg_across_groups(mtcars, mpg)
reg_across_groups(mtcars, "mpg")
# Does not work
c("mpg", "wt") %>%
map(~ reg_across_groups(mtcars, .))
I get the following error message:
Error in eval(predvars, data, env) : object '.' not found

Answer 1

We can convert to symbol with ensym
library(dplyr)
library(tidyr)
library(purrr)
reg_across_groups <- function(df, var) {
var <- ensym(var)
df <- df %>%
nest(data = -c("cyl"))
model_formula <- formula(paste0(var, "~ hp"))
df %>%
dplyr::mutate(model = purrr::map(data, ~lm(model_formula, data = .x)))
}
c("mpg", "wt") %>%
map(~ reg_across_groups(mtcars, !!.x))
#[[1]]
# A tibble: 3 x 3
# cyl data model
# <dbl> <list<df[,10]>> <list>
#1 6 [7 × 10] <lm>
#2 4 [11 × 10] <lm>
#3 8 [14 × 10] <lm>
#[[2]]
# A tibble: 3 x 3
# cyl data model
# <dbl> <list<df[,10]>> <list>
#1 6 [7 × 10] <lm>
#2 4 [11 × 10] <lm>
#3 8 [14 × 10] <lm>
reg_across_groups(mtcars, mpg)
# A tibble: 3 x 3
# cyl data model
# <dbl> <list<df[,10]>> <list>
#1 6 [7 × 10] <lm>
#2 4 [11 × 10] <lm>
#3 8 [14 × 10] <lm>
reg_across_groups(mtcars, "mpg")
# A tibble: 3 x 3
# cyl data model
# <dbl> <list<df[,10]>> <list>
#1 6 [7 × 10] <lm>
#2 4 [11 × 10] <lm>
#3 8 [14 × 10] <lm>