How scores are calculated within the step_select_linear function in the colino package?

Package development
1

I thought that the scores of the variables returned by step_select_linear with a binary outcome were the coefficients of the logistic regression model, but that is not the case.

library(recipes)
library(parsnip)
library(colino)

# load the example iris dataset
data(cells, package = "modeldata")

# create a preprocessing recipe
rec <-
 recipe(class ~ ., data = cells[, -1]) %>%
 step_select_linear(
   all_predictors(),
   outcome = "class",
   threshold = 0.9
 )

prepped <- prep(rec, new_data = cells[, -1])
prepped
step_1 <- prepped$steps[[1]]
step_1$scores %>% arrange(desc(score))
glm(class ~ inten_cooc_max_ch_3, data=cells, family=binomial )
2

I thought that the scores of the variables returned specification of a recipe step that selects a subset of predictors based on the ranking of the magnitude of coefficients. The goal of colino is to provide supervised feature selection steps to be used with the tidymodels recipes package. of a recipe step that selects a subset of predictors cabernet sauvignon wine as part of a regression model based

3

Yes, that's true, that's Colino's goal. But I like to know the details to understand exactly what each method provides, and I don't fully understand what these scores are, because I don't see them as regression coefficients.