Fix sorting in Prediction type for multiclass classification and add stronger tests #213
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
…d some extra tests
…multiclass classification
Codecov Report
@@ Coverage Diff @@
## master #213 +/- ##
==========================================
+ Coverage 86.36% 86.37% +0.01%
==========================================
Files 310 310
Lines 10136 10137 +1
Branches 351 548 +197
==========================================
+ Hits 8754 8756 +2
+ Misses 1382 1381 -1
Continue to review full report at Codecov.
|
tovbinm
reviewed
Feb 4, 2019
|
|
||
| // Need to make sure we sort the keys by their final index, which comes after an underscore in the apply function | ||
| private def keysStartsWith(name: String): Array[String] = value.keys.filter(_.startsWith(name)).toArray | ||
| .sortBy(s => s.substring(s.lastIndexOf("_") + 1).toInt) |
tovbinm
reviewed
Feb 4, 2019
| val numClasses = 5 | ||
| val numRows = 10 | ||
| val vectors = Seq.fill[OPVector](numRows)(Array.fill(numClasses)(4.2).toOPVector) | ||
| println(s"vectors: $vectors") |
|
Sorting condition did not make it? |
tovbinm
approved these changes
Feb 5, 2019
gerashegalov
reviewed
Feb 6, 2019
|
|
||
| // Need to make sure we sort the keys by their final index, which comes after an underscore in the apply function | ||
| private def keysStartsWith(name: String): Array[String] = value.keys.filter(_.startsWith(name)).toArray | ||
| .sortBy(_.split('_').last.toInt) |
There was a problem hiding this comment.
depending how wide the keys are, but it would be more efficient to scan keys backwards to look for _ with lastIndexOf to map to a substring. It would also produce less transient garbage: a single string instead of an array.
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.
Related issues
None
Describe the proposed solution
Sorts keys by their final index - the actual index corresponding to the class, rather than the default string sort which jumbles classes together. Also adds many tests relevant to multiclass classification and threshold metrics.
Describe alternatives you've considered
N/A
Additional context
Previous threshold metrics (only those, not the metrics from Spark itself) were out of order when there were more that 10 classes due to this issue.