Simon P. Couch comments

Results 165 comments of


                                            Simon P. Couch

Utility to immediately provide shareable link to data

Not at all a comprehensive solution, but I put together a package this morning to address part of the "local data frame" "csv on Google Drive" workflow that @jennybc and...

Account for possibility of custom objective function in XGBoost `boost_tree()`

Related to #774.

parsnip doesn't patch `params` argument with `xgboost` engine in `boost_tree`

Thanks for the issue and reprex! This was a hard one to track down, but your inclination that this is a xgboost + parsnip issue specifically was correct. A slightly...

parsnip doesn't patch `params` argument with `xgboost` engine in `boost_tree`

There's some related documentation in `?details_boost_tree_xgboost`, but leaves some to be said. At the moment, you can pass a different loss as `objective` to `set_engine`, but this doesn't allow interfacing...

Feature Request: ability to pass dataframe to `validation` argument of `xgboost`

Thanks for the issue! This is an interesting idea and one that we ought to consider. xgboost and lightgbm's interfaces for validation sets allow for a lot of user control,...

More details of formula usage in mgcv engine docs when using workflow

A relevant Community post with reprex: https://community.rstudio.com/t/error-in-fit-xy-with-gam-model/143065

Internals bug? `will_make_matrix()` returns `False` when given a matrix

I believe `will_make_matrix` may mean "is this a thing that's _not_ a matrix that will be converted to one," and that first `if (is.matrix) return(FALSE)` line makes me think that...

Internals bug? `will_make_matrix()` returns `False` when given a matrix

Given that we've confirmed that the function, as it's currently scoped, works as intended, and we've closed the PR that prompted the proposed extension in scope, I'm going to go...

Allow mtry to accept a proportion of features, maybe via an mtry_prop parameter?

Our docs will be clarified here on merge of #734. :)

Issue with `fit_xy()` in presence of missing values

Just chiming in that Julia's explanation here seems spot on + equally concerning to me. From my understanding, parsnip silently drops incomplete cases before handing off data to modeling functions...