Use pipe operator into `expss::uselabels()`?

Question

I am trying to pipe into expss::uselabels().

A simple replicable example of what I'm trying to do (without the pipe), would be a labelled lm() model:

library(tidyverse)
library(expss)

df <- mtcars

df <-  apply_labels(df,
                  cyl =  "Number of Cylinders",
                  disp = "Displacement")

fit_1 <- df %>% 
  use_labels(lm(formula = mpg ~ disp + cyl))
summary(fit_1)

which gives labelled coefficients in the lm output:

# > Coefficients:
#>                       Estimate Std. Error t value Pr(>|t|)    
#> (Intercept)           34.66099    2.54700  13.609 4.02e-14 ***
#> Displacement          -0.02058    0.01026  -2.007   0.0542 .  
#> `Number of Cylinders` -1.58728    0.71184  -2.230   0.0337 *

My questions: can I first take an lm() model and then pipe into use_labels()? I've tried below, but I must be refering to the two paramaters incorrectly.

fit_1<- df %>% 
  lm(formula = mpg ~ disp + cyl) %>% 
  use_labels(data = .x, expr = .y)

Looking at [the source](https://github.com/gdemin/expss/blob/ac944867320c7342e46fc8c7a30ebdd0bfdc5075/R/compute.R#L409-L411), it uses `substitute()` to delay evaluation of `expr`, which I don't think will work with a pipe because the left hand side is always evaluated before the right hand side. Your last pipeline wouldn't work regardless, because `df` is gone by the time you get to `use_labels()`, and `.x` and `.y` aren't anything; the only special variable with pipes is `.` which is the evaluated LHS being passed in and used for specifying what parameter it should be passed to. — alistaire, Jun 21 '20 at 04:06
You could use magrittrs `%T>%` pipe , but then you would have to call your model twice: `fit_3 <- df %T>% lm(mpg ~ disp + cyl, .) %>% use_labels(., lm(mpg ~ disp + cyl, .))` — TimTeaFan, Jun 21 '20 at 11:35

score 1 · Answer 1 · answered Jun 21 '20 at 14:14

use_labels works in a very simple and straightforward way. It just replaces in the expression all variables names with their labels. Variables are searched in the first argument (data.frame). As @alistaire already said all this actions are performed before the evaluation of the supplied expression, e. g. before calculating result of the lm(formula = mpg ~ disp + cyl). So answer is on your question is 'No'. You cannot apply use_labels on the already calculated result.

Use pipe operator into `expss::uselabels()`?

1 Answers1