[SPARK-10482] [ML] Add Python interface for ml.CountVectorizer

[hhbyyh]

jira: https://issues.apache.org/jira/browse/SPARK-10482

Add Python interface for feature transformer: ml.CountVectorizer

[SparkQA]

Test build #42112 has finished for PR 8650 at commit 0f1fa34.

• This patch fails Python style tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #42122 has finished for PR 8650 at commit d22ba5a.

• This patch fails Python style tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #42125 has finished for PR 8650 at commit dd0e933.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • class CountVectorizer(JavaEstimator, HasInputCol, HasOutputCol):
  • class CountVectorizerModel(JavaModel):

[mengxr]

remove sentenceData =

[mengxr]

LGTM except some minor issues

[holdenk]

This seems to do the same work as the outstanding PR #8561

[mengxr]

@holdenk Yes, I just noticed it. Could you merge some changes in this PR into yours? I think the doctest from @hhbyyh is better and the default values are specified correctly in this PR. I will make a pass after.

@hhbyyh Since this duplicates #8561, do you mind closing this PR? You can check opening PRs at https://spark-prs.appspot.com/#mllib.

[holdenk]

Ok, I'll merge in the doc tests.

[hhbyyh]

@mengxr Sorry for the extra effort during review.