Automatically predict NA for rows w/ NAs and learners that don't support missings #2099
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mb706
commented
Dec 9, 2017
| @@ -362,7 +362,7 @@ plotHyperParsEffect = function(hyperpars.effect.data, x = NULL, y = NULL, | |||
| regr.task = makeRegrTask(id = "interp", data = d.run[, c(x, y, z)], | |||
| target = z) | |||
| mod = train(lrn, regr.task) | |||
| prediction = predict(mod, newdata = grid) | |||
| prediction = predict(mod, newdata = grid[c(x, y)]) | |||
larskotthoff
requested changes
Dec 10, 2017
tests/testthat/test_base_predict.R
Outdated
| @@ -144,3 +144,11 @@ test_that("predict works with data.table as newdata", { | |||
| expect_warning(predict(mod, newdata = data.table(iris)), regexp = "Provided data for prediction is not a pure data.frame but from class data.table, hence it will be converted.") | |||
| }) | |||
|
|
|||
| test_that("predict with NA rows for learners that don't support missings automatically returns NA", { | |||
| mod = train("classif.knn", pid.task) | |||
There was a problem hiding this comment.
Could you also add a test for the original random forest problem please?
R/predictLearner.R
Outdated
|
|
||
| removeNALines = function(newdata) { | ||
| namat = is.na(newdata) | ||
| if (!any(vlapply(namat, any))) { |
There was a problem hiding this comment.
Is this check necessary? As far as I can see the code after would do the right thing in this case as well.
There was a problem hiding this comment.
I was wrong about what format the return of is.na(data.frame) has, I'll drop this part.
larskotthoff
approved these changes
Dec 10, 2017
|
Thanks, merging. |
zmjones
pushed a commit
that referenced
this pull request
Dec 19, 2017
…ort missings (#2099) * predict NA if learner doesn't support that * adding test * drop = FALSE * bugfix * using old prediction as fallback when all rows are NA * implementing @larskotthoff's suggestions
This was referenced Jun 6, 2019
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
A comprehensive fix for the larger issue around #1515, this does what I described in my comment to #2068:
If the learner doesn't support 'missings', the rows containing missings are stripped, and
NAs are added to the prediction in their place. This has two weaknesses:"missings"to mean support for missings in the prediction data.NA, this falls back to the old prediction mode (and possibly creates an error if the Learner doesn't silently ignoreNAs). A more thorough implementation could create the matrix / vector ofNAs of appropriate type without callingpredictLearnerat all.