[SPARK-6256] [MLlib] MLlib Python API parity check for regression

[yanboliang]

MLlib Python API parity check for Regression, major disparities need to be added for Python list following:

LinearRegressionWithSGD
    setValidateData
LassoWithSGD
    setIntercept
    setValidateData
RidgeRegressionWithSGD
    setIntercept
    setValidateData

setFeatureScaling is mllib private function which is not needed to expose in pyspark.

[SparkQA]

Test build #28509 has started for PR 4997 at commit 2dff3df.

• This patch merges cleanly.

[SparkQA]

Test build #28509 has finished for PR 4997 at commit 2dff3df.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/28509/
Test PASSed.

[yanboliang]

@jkbradley @mengxr Can you review this patch?

[mengxr]

@yanboliang setFeatureScaling is not a public method. We were a little hesitated to expose it. Shall we only add validateData in this PR?

[mengxr]

Use builder pattern.

[jkbradley]

Let's also add setIntercept. Also, in addition to setFeatureScaling being private, we do not need to expose optimizer.setUpdater for the 2 algorithms you listed because they have fixed updaters they should use (corresponding to the regularization they use).

[SparkQA]

Test build #28978 has started for PR 4997 at commit de5ecbc.

• This patch merges cleanly.

[SparkQA]

Test build #28978 has finished for PR 4997 at commit de5ecbc.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/28978/
Test PASSed.

[yanboliang]

Yes, "addIntercept" should be more clear and consistent.

[SparkQA]

Test build #29090 has started for PR 4997 at commit 1fb7b4f.

• This patch merges cleanly.

[SparkQA]

Test build #29090 has finished for PR 4997 at commit 1fb7b4f.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/29090/
Test PASSed.

[jkbradley]

I'm sorry! I got confused about "intercept," thinking it was being added to this class. We should stick with the original name ("intercept") since it's a public API change otherwise.

[jkbradley]

@yanboliang It looks fine to me, except for the intercept issue (sorry!) and for doc tests. Could you please add doc tests for LassoWithSGD, RidgeRegressionWithSGD using setIntercept, setValidateData? Thanks!

[SparkQA]

Test build #29160 has started for PR 4997 at commit 102f498.

• This patch merges cleanly.

[SparkQA]

Test build #29160 has finished for PR 4997 at commit 102f498.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/29160/
Test PASSed.

[jkbradley]

@yanboliang LGTM merging into master
Thanks!