Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[SPARK-9911] [DOC] [ML] Update Userguide for Evaluator #8304

Closed
wants to merge 1 commit into from

Conversation

MechCoder
Copy link
Contributor

I added a small note about the different types of evaluator and the metrics used.

@MechCoder
Copy link
Contributor Author

@mengxr

I thought it is unnecessary to add a separate guide for evaluators and hence added a note within the existing example.

@SparkQA
Copy link

SparkQA commented Aug 19, 2015

Test build #41221 has finished for PR 8304 at commit 4a4d7a6.

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


The `Evaluator` can be a [`RegressionEvaluator`](api/scala/index.html#org.apache.spark.ml.RegressionEvaluator)
for regression problems, a [`BinaryClassificationEvaluator`](api/scala/index.html#org.apache.spark.ml.BinaryClassificationEvaluator)
for binary data or a [`MultiClassClassificationEvaluator`](api/scala/index.html#org.apache.spark.ml.MultiClassClassificationEvaluator)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: Oxford comma ("...binary data, or a...")

@feynmanliang
Copy link
Contributor

+1 for @mengxr 's suggestion on JIRA to break this example into a separate ml-evaluation.md section; the examples in ml-guide are becoming monolithic and breaking another section will allow us to better document the features available in evaluators (e.g. all the metrics supported, whether to maximize or minimize the metric a la #8290)

@MechCoder
Copy link
Contributor Author

That sounds like a great idea. But do you have better suggestions for an example? I would have to use a cross-validator and tune it anyway and I'm afraid there will be code repetition.

@feynmanliang
Copy link
Contributor

I don't think we need code examples for every possible combination, otherwise combinatorial complexity is going to bite us hard.

We can:

  • List all the evaluators and their features
  • List the (currently two?) validators, CV and TrainTestSplit, and their features
  • Provide the current code example going through with the binary classification metric

@MechCoder
Copy link
Contributor Author

otherwise combinatorial complexity is going to bite us hard.

lol. That is not what I had meant. Sorry I should be more clear from here on.

In other words I was asking if it would be sufficient to move the Model Selection Example to the new ml-evalation.md or if you had anything else in mind?

@feynmanliang
Copy link
Contributor

Oh, sorry!

I think moving the example and providing some guidance about how to choose evaluators/validators is sufficient for now.

@MechCoder
Copy link
Contributor Author

ping @jkbradley ?

@jkbradley
Copy link
Member

+1 for "moving the example and providing some guidance about how to choose evaluators/validators"

Something simple, but separate, which we can build upon later on.

@mengxr
Copy link
Contributor

mengxr commented Aug 28, 2015

@MechCoder is busy this week. I will make a PR based on this.

asfgit pushed a commit that referenced this pull request Aug 28, 2015
I added a small note about the different types of evaluator and the metrics used.

Author: MechCoder <manojkumarsivaraj334@gmail.com>

Closes #8304 from MechCoder/multiclass_evaluator.

(cherry picked from commit 30734d4)
Signed-off-by: Xiangrui Meng <meng@databricks.com>
@mengxr
Copy link
Contributor

mengxr commented Aug 28, 2015

Since all comments are minor (or beyond this PR), I'm going to merge this into master and branch-1.5 first and then send another PR to fix issues here and some others.

@asfgit asfgit closed this in 30734d4 Aug 28, 2015
@MechCoder MechCoder deleted the multiclass_evaluator branch August 28, 2015 14:28
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
5 participants