add butcher methods for mixOmics output

[simonpcouch]

These methods do a good job of being selective about what they bring along from the enclosing env, and do make use of most all of the heavier bits of inputted data that they retain at predict time—not a lot of memory released, in the end.

mixo_plsda objects are a subclass of mixo_pls, so don’t need their own methods!

The mixOmics package is hosted on Bioconductor rather than CRAN, so this PR doesn’t introduce mixOmics as a dependency. This also means that mixOmics methods won’t be tested on CRAN or CI.

library(butcher)
library(mixOmics)
#> Loading required package: MASS
#> Loading required package: lattice
#> Loading required package: ggplot2
#> 
#> Loaded mixOmics 6.20.0
#> Thank you for using mixOmics!
#> Tutorials: http://mixomics.org
#> Bookdown vignette: https://mixomicsteam.github.io/Bookdown
#> Questions, issues: Follow the prompts at http://mixomics.org/contact-us
#> Cite us:  citation('mixOmics')

fit_mod <- function() {
  boop <- runif(1e6)
  pls(matrix(rnorm(2e4), ncol = 2), rnorm(1e4), mode = "classic")
}

mod_fit <- fit_mod()
mod_res <- butcher(mod_fit)

weigh(mod_fit)
#> # A tibble: 24 × 2
#>    object             size
#>    <chr>             <dbl>
#>  1 X              0.801   
#>  2 variates.X     0.801   
#>  3 variates.Y     0.801   
#>  4 Y              0.721   
#>  5 names.sample   0.640   
#>  6 input.X        0.161   
#>  7 call           0.00129 
#>  8 loadings.X     0.000776
#>  9 loadings.Y     0.000696
#> 10 loadings.star1 0.0006  
#> # … with 14 more rows
weigh(mod_res)
#> # A tibble: 24 × 2
#>    object              size
#>    <chr>              <dbl>
#>  1 X               0.801   
#>  2 variates.X      0.801   
#>  3 variates.Y      0.801   
#>  4 Y               0.721   
#>  5 loadings.X      0.000776
#>  6 loadings.Y      0.000696
#>  7 loadings.star1  0.0006  
#>  8 mat.c           0.0006  
#>  9 loadings.star2  0.00052 
#> 10 prop_expl_var.X 0.000352
#> # … with 14 more rows

new_data <- matrix(1:2, ncol = 2)
colnames(new_data) <- c("X1", "X2")

testthat::expect_equal(
  purrr::discard_at(predict(mod_fit, new_data), "call"),
  purrr::discard_at(predict(mod_res, new_data), "call")
)

^{Created on 2023-01-23 with reprex v2.0.2}

Related to #234.

[juliasilge]

Do you think it would be worth it to install the package into the check action, so the tests get run on CI? Or not very worthwhile?

[simonpcouch]

I think that's a good idea. :) Added in 0d607c5.

EDIT: Seeing those failures, trying to debug🤔

[juliasilge]

This is great -- thank you again! 👨‍🎤

[github-actions]

This pull request has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.