[SPARK-7633] [MLlib] [PySpark] Python bindings for StreamingLogisticRegressionwithSGD

[MechCoder]

Add Python bindings to StreamingLogisticRegressionwithSGD.

No Java wrappers are needed as models are updated directly using train.

[SparkQA]

Test build #35005 has finished for PR 6849 at commit 7bd6a46.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class StreamingLogisticRegressionWithSGD(object):

[SparkQA]

Test build #35002 timed out for PR 6849 at commit 0e2fe86 after a configured wait of 175m.

[MechCoder]

jenkins retest this please

[MechCoder]

cc @freeman-lab :P

[SparkQA]

Test build #35019 has finished for PR 6849 at commit 7bd6a46.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class StreamingLogisticRegressionWithSGD(object):

[SparkQA]

Test build #35358 has finished for PR 6849 at commit 5a94db0.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class StreamingLogisticRegressionWithSGD(object):

[MechCoder]

@mengxr Could you review this? This should be the next one to go as I have refactored the code.

[SparkQA]

Test build #35363 has finished for PR 6849 at commit 19c0f25.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class StreamingLinearAlgorithm(object):
  • class StreamingLogisticRegressionWithSGD(StreamingLinearAlgorithm):

[mengxr]

organize imports

[mengxr]

@MechCoder I need to make another pass on the code and tests after your update. Please make sure the tests are the same as the ones on the Scala side.

[MechCoder]

@mengxr Thanks for the updates! I confirm that the tests are very similar

[SparkQA]

Test build #35688 has finished for PR 6849 at commit bdab425.

• This patch fails MiMa tests.
• This patch does not merge cleanly.
• This patch adds the following public classes (experimental):
  • class StreamingLinearAlgorithm(object):
  • class StreamingLogisticRegressionWithSGD(StreamingLinearAlgorithm):

[SparkQA]

Test build #35689 has finished for PR 6849 at commit 5c4305c.

• This patch fails PySpark unit tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class StreamingLinearAlgorithm(object):
  • class StreamingLogisticRegressionWithSGD(StreamingLinearAlgorithm):

[MechCoder]

I have rerun the tests. (Strange that the tests that failed in the previous run were related to StreamingKMeans)

[SparkQA]

Test build #35699 has finished for PR 6849 at commit b4376a5.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class StreamingLinearAlgorithm(object):
  • class StreamingLogisticRegressionWithSGD(StreamingLinearAlgorithm):

[AmplabJenkins]

Merged build finished. Test FAILed.

[mengxr]

@MechCoder The streaming k-means tests seem to be flaky. Could you run it multiple times on your local?

[MechCoder]

Sure, btw the last failure was some random error.

[MechCoder]

@mengxr Could you give another pass? I can update tomorrow, there is some problem with my ubunt.

[mengxr]

LGTM. Merged into master. I think the streaming k-means test might be flaky, which is usually hard to debug. Let's watch Jenkins master build: https://amplab.cs.berkeley.edu/jenkins/job/Spark-Master-SBT and see if there are failures.