[SPARK-16750] [ML] Fix GaussianMixture training failed due to feature column type mistake #14378
Conversation
This file contains hidden or 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
|
Test build #62917 has finished for PR 14378 at commit
|
Contributor
Author
|
cc @srowen |
| @Since("2.0.0") | ||
| override def fit(dataset: Dataset[_]): MinMaxScalerModel = { | ||
| transformSchema(dataset.schema, logging = true) | ||
| transformSchema(dataset.schema) |
Member
There was a problem hiding this comment.
Just wondering why you remove the logging flag here? I know it just adds some debug logging, but there are other similar calls that still have it set to true, should those be removed also?
Contributor
Author
There was a problem hiding this comment.
It's a good question. Since MinMaxScaler override transformSchema with no argument logging, we should use that one rather than the function in the base class.
Contributor
There was a problem hiding this comment.
Seems the transformSchema(schema: StructType, logging: Boolean) method of the base class PipelineStage would call the the overloaded transformSchema method without the logging param:
Contributor
Author
There was a problem hiding this comment.
Thanks for your remind, updated the PR.
Member
|
I just had a minor question, but LGTM |
|
Test build #62970 has finished for PR 14378 at commit
|
Member
|
Seems reasonable to me. |
asfgit
pushed a commit
that referenced
this pull request
Jul 29, 2016
…column type mistake ## What changes were proposed in this pull request? ML ```GaussianMixture``` training failed due to feature column type mistake. The feature column type should be ```ml.linalg.VectorUDT``` but got ```mllib.linalg.VectorUDT``` by mistake. See [SPARK-16750](https://issues.apache.org/jira/browse/SPARK-16750) for how to reproduce this bug. Why the unit tests did not complain this errors? Because some estimators/transformers missed calling ```transformSchema(dataset.schema)``` firstly during ```fit``` or ```transform```. I will also add this function to all estimators/transformers who missed in this PR. ## How was this patch tested? No new tests, should pass existing ones. Author: Yanbo Liang <ybliang8@gmail.com> Closes #14378 from yanboliang/spark-16750. (cherry picked from commit 0557a45) Signed-off-by: Sean Owen <sowen@cloudera.com>
Member
|
Merged to master/2.0 |
asfgit
pushed a commit
that referenced
this pull request
Aug 5, 2016
…ctorAssembler and fix failed tests. ## What changes were proposed in this pull request? This is follow-up for #14378. When we add ```transformSchema``` for all estimators and transformers, I found there are tests failed for ```StringIndexer``` and ```VectorAssembler```. So I moved these parts of work separately in this PR, to make it more clear to review. The corresponding tests should throw ```IllegalArgumentException``` at schema validation period after we add ```transformSchema```. It's efficient that to throw exception at the start of ```fit``` or ```transform``` rather than during the process. ## How was this patch tested? Modified unit tests. Author: Yanbo Liang <ybliang8@gmail.com> Closes #14455 from yanboliang/transformSchema. (cherry picked from commit 6cbde33) Signed-off-by: Sean Owen <sowen@cloudera.com>
ghost
pushed a commit
to dbtsai/spark
that referenced
this pull request
Aug 5, 2016
…ctorAssembler and fix failed tests. ## What changes were proposed in this pull request? This is follow-up for apache#14378. When we add ```transformSchema``` for all estimators and transformers, I found there are tests failed for ```StringIndexer``` and ```VectorAssembler```. So I moved these parts of work separately in this PR, to make it more clear to review. The corresponding tests should throw ```IllegalArgumentException``` at schema validation period after we add ```transformSchema```. It's efficient that to throw exception at the start of ```fit``` or ```transform``` rather than during the process. ## How was this patch tested? Modified unit tests. Author: Yanbo Liang <ybliang8@gmail.com> Closes apache#14455 from yanboliang/transformSchema.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
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.
What changes were proposed in this pull request?
ML
GaussianMixturetraining failed due to feature column type mistake. The feature column type should beml.linalg.VectorUDTbut gotmllib.linalg.VectorUDTby mistake.See SPARK-16750 for how to reproduce this bug.
Why the unit tests did not complain this errors? Because some estimators/transformers missed calling
transformSchema(dataset.schema)firstly duringfitortransform. I will also add this function to all estimators/transformers (exceptStringIndexerandVectorAssemblerwhich will be addressed later in a follow up PR) who missed in this PR.How was this patch tested?
No new tests, should pass existing ones.