Scaler and descaler transformers #223
Merged
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
Codecov Report
@@ Coverage Diff @@
## master #223 +/- ##
==========================================
+ Coverage 86.36% 86.39% +0.03%
==========================================
Files 310 312 +2
Lines 10137 10182 +45
Branches 347 333 -14
==========================================
+ Hits 8755 8797 +42
- Misses 1382 1385 +3
Continue to review full report at Codecov.
|
tovbinm
reviewed
Feb 11, 2019
features/src/main/scala/com/salesforce/op/stages/impl/feature/ScalingType.scala
Show resolved
Hide resolved
core/src/main/scala/com/salesforce/op/stages/impl/feature/ScalerTransformer.scala
Outdated
Show resolved
Hide resolved
core/src/test/scala/com/salesforce/op/stages/impl/feature/DescalerTransformerTest.scala
Outdated
Show resolved
Hide resolved
core/src/test/scala/com/salesforce/op/stages/impl/feature/LinearScalerTest.scala
Outdated
Show resolved
Hide resolved
core/src/test/scala/com/salesforce/op/stages/impl/feature/LinearScalerTest.scala
Show resolved
Hide resolved
| import scala.util.{Failure, Success} | ||
|
|
||
| @RunWith(classOf[JUnitRunner]) | ||
| class ScalerMetadataTest extends FlatSpec with TestSparkContext{ |
There was a problem hiding this comment.
missing a space as well TestSparkContext {
core/src/test/scala/com/salesforce/op/stages/impl/feature/ScalerTest.scala
Outdated
Show resolved
Hide resolved
Merged
Merged
|
Thanks for the contribution! Before we can merge this, we need @ericwayman to sign the Salesforce.com Contributor License Agreement. |
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Outdated suggestions cannot be applied.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
Often in regression use cases the response variable is not normally distributed. As a data scientist training regression models I'd like generic transformers to easily scale the response variable during training time, but to also descale the predictions (at scoring time) so predictions are returned in the original scale.
Proposed Solution
A general framework for scaling and descaling the response feature that allows developers to easily add support for new feature scaling functions with minimal boiler plate code and without having to write new code to handle reading and writing metadata.
Each family of scaling functions is represented by a case class which extends the
Scalercase class.All supported families of scaling functions are stored in the
ScalingTypeenum.The
ScalerTransformerconstructs a transformer that scales features fromscalingTypeandscalingArgs. ThescalingTypeandscalingArgsare stored in the metadata of this feature.The
DescalerTransformertakes in a feature for descaling and a scaled feature containing the metadata for constructing the scaling inverse. This allows for predictions (in the scaled domain) to be descaled at scoring time.