data_select(): add option to evaluate a function passed to `select`

[DominiqueMakowski]

data_select(): add option to evaluate a function passed to `select`

[etiennebacher]

You can evaluate the function beforehand and use its output:

library(datawizard)

model <- lm(Petal.Length ~ Petal.Width, data=iris)

cols <- insight::find_predictors(model, effects = "fixed", flatten = TRUE)

head(data_select(iris, cols))
#>   Petal.Width
#> 1         0.2
#> 2         0.2
#> 3         0.2
#> 4         0.2
#> 5         0.2
#> 6         0.4

There are already many arguments in functions that use select helpers so I don't think adding force_eval would be worth it.

[etiennebacher]

Also, in this case it would be easy but if you want to do this in a custom function then evaluating can be tricky:

foo <- function(effects) {
  data_select(iris, insight::find_predictors(model, effects = effects, flatten = TRUE))
}

Evaluating insight::find_predictors(model, effects = effects, flatten = TRUE) would error because we need to get effects first

[DominiqueMakowski]

or maybe a safeguard, like try to running the caught function on the data (as it is primarily expected) and if it fails try to run the expression on its own and if it fails again then throw an error

[etiennebacher]

@strengejacke @IndrajeetPatil what do you think?

[strengejacke]

#414 should fix this issue. (let's wait for checks...)

[strengejacke]

library(datawizard)
model <- lm(Petal.Length ~ Petal.Width, data=iris)

head(data_select(iris, insight::find_predictors(model, effects = "fixed", flatten = TRUE)))
#>   Petal.Width
#> 1         0.2
#> 2         0.2
#> 3         0.2
#> 4         0.2
#> 5         0.2
#> 6         0.4

effects <- "fixed"
head(data_select(iris, insight::find_predictors(model, effects = effects, flatten = TRUE)))
#>   Petal.Width
#> 1         0.2
#> 2         0.2
#> 3         0.2
#> 4         0.2
#> 5         0.2
#> 6         0.4

^{Created on 2023-05-02 with reprex v2.0.2}