improve error messages #80
Conversation
| @@ -89,6 +89,7 @@ | |||
| #' cart_pca_bag | |||
| #' } | |||
| #' @export | |||
| #' @include validate.R | |||
| } | ||
|
|
||
| # TODO extend to use mode from model object(s) |
There was a problem hiding this comment.
We could eventually extend baguette to censored regression models. There is a lot of code that looks at the data to get the mode or assumes that they are either regression or classification.
| } | ||
|
|
||
| if (!is.na(msg)) { | ||
| # escape any brackets in the error message |
There was a problem hiding this comment.
A long way to get there but this should make much much nicer error messages when models fail
| Code | ||
| bagger(Sepal.Length ~ ., data = iris, times = 2L, base_model = "C5.0") | ||
| Condition | ||
| Error in `validate_outcomes_are_factors()`: |
There was a problem hiding this comment.
All of our snapshots are going to change once these hardhat functions can take calls
There was a problem hiding this comment.
Looking good!
by looking at the snapshots, i see some calls aren't passed through to hardhat::validate_outcomes_are_factors() and
validate_outcomes_are_univariate(). these don't yet have call arguments so we have to wait for {hardhat} for them
there is also a call to stop() on line 100 of R/bagger.R that needs to be converted. Wasn't able to tag directly as it is "outside of github scope"
| expect_snapshot({ | ||
| set.seed(459394) | ||
| bagger(a ~ ., data = bad_iris, base_model = "CART", times = 3) | ||
| }, | ||
| error = TRUE | ||
| ) |
There was a problem hiding this comment.
This snapshot bubbles up the issue that check_for_disaster() doesn't pass calls around
There was a problem hiding this comment.
at one point in time. ideally this should say Error in bagger(): not Error in check_for_disaster():
| } | ||
|
|
||
| integer_B <- function(B) { | ||
| if (is.numeric(B) & !is.integer(B)) { |
There was a problem hiding this comment.
| if (is.numeric(B) & !is.integer(B)) { | |
| if (is.double(B)) { |
| msg <- gsub("(\\{)", "\\1\\1", msg) | ||
| msg <- gsub("(\\})", "\\1\\1", msg) | ||
| msg <- cli::format_error(msg) |
There was a problem hiding this comment.
as per r-lib/cli#735 (comment)
| msg <- gsub("(\\{)", "\\1\\1", msg) | |
| msg <- gsub("(\\})", "\\1\\1", msg) | |
| msg <- cli::format_error(msg) | |
| msg <- cli::format_error("{msg}") |
| rpart, | ||
| rsample, | ||
| tibble, | ||
| tidyr, | ||
| utils, | ||
| withr | ||
| Suggests: | ||
| AmesHousing, |
There was a problem hiding this comment.
i'm going to assume that this is leftover from a long time ago, and that we don't need it since we have modeldata
This needed sooo much upkeep that it was difficult to separate the different modes of error upgrades.
This PR adds type checkers, refines cli messages, routes calls to the user-facing function, and cleans up a bunch of dplyr/purrr calls that lack namespaces.