add toggle_sparsity()
#281
Conversation
R/sparsevctrs.R
Outdated
| if (is.null(model) || model == "ranger") { | ||
| return("no") | ||
| } |
There was a problem hiding this comment.
since ranger performs identically for both sparse and dense data, we turn off automatically
There was a problem hiding this comment.
Loving where this is going! Finding the inlined model fit very satisfying😆
I know that we're planning on testing all of the unique combinations of these code paths in extratests, but could we test toggle_sparsity() by itself here as well as one example workflow fit where toggle_sparsity() returns "yes"?
Co-authored-by: Simon P. Couch <[email protected]>
tests/testthat/test-sparsevctrs.R
Outdated
|
|
||
| data("ames", package = "modeldata") | ||
|
|
||
| tree_spec <- parsnip::boost_tree("regression", "xgboost") |
There was a problem hiding this comment.
Seeing:
══ Failed tests ════════════════════════════════════════════════════════════════
── Failure ('test-sparsevctrs.R:294:3'): toggle_sparsity doesn't break fit ─────
Expected `fit(wf_spec, ames)` to run without any errors.
i Actually got a <rlang_error> with text:
Please install the xgboost package to use this engine.
in tests.
A few of these read skip_if_not_installed("glmnet") but then use an xgboost model spec.
|
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. |
to close #271
I think this is the last puzzle piece to make sparsity useful in tidymodels.
This will toggle the sparsity creation of recipes based on whether or not we estimate it will be useful or not.