[SPARK-9768] [PySpark] [ML] Add Python API and user guide for ml.feature.ElementwiseProduct

[yanboliang]

Add Python API, user guide and example for ml.feature.ElementwiseProduct.

[yanboliang]

Jenkins, test this please.

[SparkQA]

Test build #40270 has finished for PR 8061 at commit 48239ab.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class ElementwiseProduct(JavaTransformer, HasInputCol, HasOutputCol):

[jkbradley]

style: indent to match parenthesis, as in PySpark internal code

[jkbradley]

Thanks for the PR! Just small comments.

[SparkQA]

Test build #40628 has finished for PR 8061 at commit 2ff657f.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class ElementwiseProduct(JavaTransformer, HasInputCol, HasOutputCol):

[yanboliang]

Jenkins, test this please.

[jkbradley]

LGTM once merge conflicts are fixed and tests pass

[SparkQA]

Test build #40713 timed out for PR 8061 at commit 5c87ec3 after a configured wait of 175m.

[SparkQA]

Test build #40738 has finished for PR 8061 at commit d31ed8a.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • trait Identifiable
  • class VectorUDT extends UserDefinedType[Vector]
  • class ElementwiseProduct(JavaTransformer, HasInputCol, HasOutputCol):

[jkbradley]

New though: I'm going to quickly test to see if the scalingVec can be set as a numpy or Python array. If that works, then I'll merge it. Otherwise, it'll be nice to add.

[jkbradley]

Actually, thinking more about it, that should be handled in the wrapper (_transfer_params_to_java) and might take too long to do. Would you mind noting in the built-in Param doc for scalingVec that it must be an MLlib Vector type? After that, this should be ready.

[SparkQA]

Test build #40947 has finished for PR 8061 at commit 1dd6224.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class ElementwiseProduct(JavaTransformer, HasInputCol, HasOutputCol):

[yanboliang]

@jkbradley I have updated Param doc. I opened SPARK-10009 to track the issue of supporting Param of Vector type can be set with Python array or numpy.array. I think we can resolve it in the next release cycle.

[jkbradley]

test this please

[jkbradley]

just testing once more, then will merge

[SparkQA]

Test build #41072 has finished for PR 8061 at commit 1dd6224.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class ElementwiseProduct(JavaTransformer, HasInputCol, HasOutputCol):

[jkbradley]

Merging with master and branch-1.5