[SPARK-12159] [ML] Add user guide section for IndexToString transformer

[BenFradet]

Documentation regarding the IndexToString label transformer with code snippets in Scala/Java/Python.

[SparkQA]

Test build #47245 has finished for PR 10166 at commit 10ba98a.

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

[holdenk]

It might be useful to also document the different ways "missing" labels can be handled - what are your thoughts?

[BenFradet]

Hey @holdenk, thanks for reviewing.

Do you mean regarding StringIndexer#setHandleInvalid method? If so, yes that'd be a good addition.

However, I'm not sure if I should include it in this jira/pr or create another, input welcome.

[BenFradet]

cc @jkbradley

[holdenk]

That is what I was referring to, handling it in a follow up JIRA/PR seems ok too (just since one of the things blocking the original implementation was wanting to have it be user controllable if we allowed people to specify their own maps it seemed like good for that to also make it through to the docs).

[jkbradley]

Thanks for the PR; I'll take a look now!
@holdenk handleInvalid should be in a separate PR since it's for StringIndexer, but I agree it'd be nice to add to the docs.

[jkbradley]

Would you mind moving these to examples/ and pulling the code snippets into here using the include_example functionality? You can find examples of include_example in this .md file. This makes the examples easier to test & maintain.

[jkbradley]

Those are my only comments; the examples look good. Btw, it's OK this time, but in general, I'd recommend doing little cleanups in a separate PR. Especially when lots of docs are being merged, it's really easy to hit merge conflicts. Thanks! I'll watch for updates.

[BenFradet]

@jkbradley Thanks for reviewing, will take those comments into account.

[jkbradley]

That was a spurious test failure; I asked it to retest

[SparkQA]

Test build #2184 has finished for PR 10166 at commit cb33653.

• This patch fails Python style tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):\n * public class JavaIndexToStringExample\n

[jkbradley]

dev/lint-python should catch these issues

[jkbradley]

LGTM except for the Python style issue

[SparkQA]

Test build #47361 has finished for PR 10166 at commit 9591007.

• This patch fails Python style tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):\n * public class JavaIndexToStringExample\n

[SparkQA]

Test build #47363 has finished for PR 10166 at commit 9398743.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):\n * public class JavaIndexToStringExample\n

[jkbradley]

Merging with master and branch-1.6

Thanks for the PR!

[BenFradet]

@jkbradley Should I log a jira for completing the user guide on StringIndexer regarding the handling of missing labels @holdenk was talking about?

[jkbradley]

If you wouldn't mind, that'd be great, thanks!

[BenFradet]

will do