[SPARK-12633][Python][MLlib][DOC] Update param descriptions in regression.py

[vijaykiran]

Updates the param, in python mllib's regression.py, descriptions to be consistent. See [SPARK-11219] for
more details.

[jkbradley]

I just added a note to the parent JIRA about a formatting issue affecting all 5 PRs: [https://issues.apache.org/jira/browse/SPARK-11219?focusedCommentId=15090225&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-15090225]
Could you please check it out & ping when I should review again? Thank you!

[vijaykiran]

ping @jkbradley !

[BryanCutler]

Capitalize the first word and add a period at the end

[vijaykiran]

Do you mean "pyspark.mllib" should be "Pyspark" ? I thought package names should be left alone.

[BryanCutler]

Sorry, my bad - it is already capitalized, my font was too small I guess :). Could you just add a period after the end of the description? So should read ..."scipy.sparse column matrix)."

[BryanCutler]

Thanks @vijaykiran , just a couple minor corrections and would you mind changing the PR title to indicate this is for regression.py?

[vijaykiran]

@BryanCutler Fixed the indentation and added period.

[mengxr]

ok to test

[SparkQA]

Test build #50151 has finished for PR 10600 at commit 5feecba.

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

[BryanCutler]

we need to settle on the :Allowed values: format, also in classification.py, before we can finish this one up.

[MLnick]

@BryanCutler it looks like a version of the "allowed values" got merged in the PR for classification.py - https://github.com/apache/spark/pull/11183/files#diff-3a683f424bec70006f6c225a32d162f9R297?

[BryanCutler]

Yeah, we were discussing the format of "allowed values" in that PR here I just wanted to make sure we used the same format for the similar params here. I'm not too crazy about the extra blank lines, but Sphinx requires them to format it correctly without generating errors.

[MLnick]

@BryanCutler ok, from that comment discussion it seems the format agreed would be:

Allowed values:

- "l1" ...
- "l2" ...

As we discussed in the PR for tree.py, I prefer Supported values over Allowed values, though that would require updating it here and in classification.py (and #10598 has been merged already).

[BryanCutler]

@MLnick I don't have a preference for Supported values vs Allowed values but I would like to see it consistent throughout, so it should probably be updated in classification.py. Should I do that here, or open another issue for that?

[MLnick]

It's pretty minor so just add it here
On Fri, 26 Feb 2016 at 21:31, Bryan Cutler [email protected] wrote:

@MLnick https://github.com/MLnick I don't have a preference for Supported
values vs Allowed values but I would like to see it consistent
throughout, so it should probably be updated in classification.py. Should
I do that here, or open another issue for that?

—
Reply to this email directly or view it on GitHub
#10600 (comment).