Accessing linear model statistics inside of tidymodel

Machine Learning and Modeling
,
1

Is there a tidymodels way to extract all the statistics from a parsnip model of linear regression?

What I want is the same results as obtained with

summary(lm(mpg ~ wt, mtcars))

I could do it this way:

library(tidymodels)

fit_tidy <- linear_reg() %>%
  set_engine("lm") %>%
  fit(mpg ~ wt, mtcars)

summary(fit_tidy$fit)

But I wonder if there is a function from tidymodels to do that.

All help very much appreciated!

2

tidymodels incorporates the broom library, which contains the tidy() function

tidy(fit_tidy)
3

Thank you! This is very helpful! But, still, I need the other metrics too, such as Residual Standard error, Multiple and Adjusted R squared, and so on.

4

Maybe the glance() method from broom is probably what you want?

library(parsnip)
library(broom)

fit_tidy <- linear_reg() %>%
  set_engine("lm") %>%
  fit(mpg ~ wt, mtcars)

glance(fit_tidy)
#> # A tibble: 1 × 12
#>   r.squared adj.r.squared sigma statistic  p.value    df logLik   AIC   BIC
#>       <dbl>         <dbl> <dbl>     <dbl>    <dbl> <dbl>  <dbl> <dbl> <dbl>
#> 1     0.753         0.745  3.05      91.4 1.29e-10     1  -80.0  166.  170.
#> # … with 3 more variables: deviance <dbl>, df.residual <int>, nobs <int>

Created on 2022-10-11 by the reprex package (v2.0.1)

5

What an honor to have an answer from you, Max! I attended ML with Tidymodels workshop in Rstudio conference, and also got your book - signed!

Actually, this is not what I wanted, but it is also interesting, because it contains some metrics that the base R way do not deliver from the summary(fit) command (AIC, BIC, etc.).

I wanted the exact same output from summary(fit) because I am working on explaining one by one in a blog post.

Maybe I will add the tidymodels version and explain the other metrics as well.

Thank you very much!