Docs for get datagrid

[mattansb]

Closes #978

[mattansb]

Errors don't seem to be related to this change.

[strengejacke]

Errors don't seem to be related to this change.

@vincentarelbundock this seems to be related to easystats/parameters#1048, right?

Failing test is:

══ Failed tests ════════════════════════════════════════════════════════════════
── Failure ('test-marginaleffects.R:34:3'): marginaleffects ────────────────────
Names of parameters::model_parameters(x) ('rowid', 'Parameter', 'Comparison', 'Coefficient', 'SE', 'Statistic', 'p', 'S', 'CI', 'CI_low', 'CI_high', 'Predicted', 'Species', 'Petal.Length', 'Sepal.Width') don't match 'rowid', 'Parameter', 'Coefficient', 'SE', 'Statistic', 'p', 'S', 'CI', 'CI_low', 'CI_high', 'Predicted', 'Species', 'Petal.Length', 'Sepal.Width'

So we make this conditional on marginaleffects 1.0.0?

[strengejacke]

@mattansb The check-hard error is related to this PR:

get_datagrid.emmGrid needs examples run conditionally.
I think we just need to add quietly = TRUE here, else check_if_installed() would throw an error.

[vincentarelbundock]

Yes, I think that's right. The current main version from Github should work, in principle.

[strengejacke]

hm, strange, we're importing the current main via the Remotes field.

[strengejacke]

We don't expect the Comparison column:

packageVersion("marginaleffects")
#> [1] '0.24.0.6'

m <- lm(Sepal.Width ~ Species * Petal.Length, data = iris)

x <- marginaleffects::slopes(m,
  variables = "Petal.Length",
  newdata = insight::get_datagrid(m, by = "Species")
)

setdiff(
  colnames(parameters::model_parameters(x)),
  c(
    "rowid", "Parameter", "Coefficient", "SE", "Statistic", "p",
    "S", "CI", "CI_low", "CI_high", "Predicted", "Species",
    "Petal.Length", "Sepal.Width"
  )  
)
#> [1] "Comparison"

^{Created on 2024-12-13 with reprex v2.1.1}

[strengejacke]

Or is the Comparison column now expected? Looks like the test in parameters expects it. Strange, this test wasn't touched for months.

[vincentarelbundock]

Hmm, I don't know what is going on. Maybe these kinds of tests are too strict? We don't want a break whenever a package moves columns around, unless it is a "material" change.

[strengejacke]

Agree, I think its purpose is to check whether the s-value column was renamed into "S". will fix test.

[strengejacke]

Ah, no, that was the test in parameters :-D

But, most important: the column Comparison, is it included in the output (and expected) or not? I remember you saying that some columns should be removed in print(), but not in the underlying data frame.

[vincentarelbundock]

In the current CRAN, it is not include. In the future CRAN versions and on Github, it will be included.

[strengejacke]

ok, I fixed the text. Shouldn't be crucial, as it's not tested on CRAN anyway.

[mattansb]

Thanks for taking care of this @strengejacke 🙏