[SPARK-7574][ml][doc] User guide for OneVsRest

[harsha2010]

Including Iris Dataset (after shuffling and relabeling 3 -> 0 to confirm to 0 -> numClasses-1 labeling). Could not find an existing dataset in data/mllib for multiclass classification.

[SparkQA]

Test build #33176 has finished for PR 6296 at commit bb9dbfa.

• This patch fails Spark unit tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • [OneVsRest](api/scala/index.html#org.apache.spark.ml.classifier.OneVsRest) is an example of a machine learning reduction for performing multiclass classification given a base classifier that can perform binary classification efficiently.
  • ``OneVsRestis anEstimator` takes as base classifier instances of `Classifier` and creates a binary classification problem for each of the k classes. The classifier for class i is trained to predict whether the label is i or not, distinguishing class i from all other classes.`
  • [LIBSVM data file](http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/), parse it as an RDD ofLabeledPointand perform multiclass classification usingOneVsRest. The test error is calculated to measure the algorithm accuracy.

[SparkQA]

Test build #33179 has finished for PR 6296 at commit 13bed9c.

• This patch fails Spark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.

[jkbradley]

Please match ml-features.md:

**Table of Contents**

* This will become a table of contents (this text will be scraped).
{:toc}

[jkbradley]

I like adding the multiclass dataset.

Could you please add a Java example?

I'm testing the Scala example now...

[jkbradley]

Please indent

[harsha2010]

@jkbradley thanks for the review. I'll add the Java Example and fix some indentation issues and review comments and update the PR tonight

[SparkQA]

Test build #33180 timed out for PR 6296 at commit 4b7d1a6 after a configured wait of 150m.

[jkbradley]

Could you please remove the use of MetadataUtils? That will no longer be public because of this PR: [https://github.com//pull/6322]

[harsha2010]

Thanks for noticing this, will do

On May 21, 2015, at 5:14 PM, jkbradley notifications@github.com wrote:

Could you please remove the use of MetadataUtils? That will no longer be public because of this PR: [#6322]

—
Reply to this email directly or view it on GitHub.

[SparkQA]

Test build #33313 has finished for PR 6296 at commit c026613.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • [OneVsRest](http://en.wikipedia.org/wiki/Multiclass_classification#One-vs.-rest) is an example of a machine learning reduction for performing multiclass classification given a base classifier that can perform binary classification efficiently.
  • [LIBSVM data file](http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/), parse it as a DataFrame and perform multiclass classification usingOneVsRest. The test error is calculated to measure the algorithm accuracy.

[jkbradley]

Sorry, I should have commented before: This import is not needed.

[jkbradley]

Line too long. Can it be split? For an example, a foreach with a println inside the loop might be easiest to understand.

[jkbradley]

Could you please use DataFrame.randomSplit here too to make the Scala & Java examples as similar as possible?

[jkbradley]

@harsha2010 Thanks for adding the Java example. Just had a couple of small comments about clarity

[jkbradley]

@harsha2010 Thank you! LGTM pending tests

[SparkQA]

Test build #33346 has finished for PR 6296 at commit ebdf103.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • [OneVsRest](http://en.wikipedia.org/wiki/Multiclass_classification#One-vs.-rest) is an example of a machine learning reduction for performing multiclass classification given a base classifier that can perform binary classification efficiently.
  • [LIBSVM data file](http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/), parse it as a DataFrame and perform multiclass classification usingOneVsRest. The test error is calculated to measure the algorithm accuracy.

[SparkQA]

Test build #33347 has finished for PR 6296 at commit 2f76295.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • [OneVsRest](http://en.wikipedia.org/wiki/Multiclass_classification#One-vs.-rest) is an example of a machine learning reduction for performing multiclass classification given a base classifier that can perform binary classification efficiently.
  • [Iris dataset](http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/multiclass/iris.scale), parse it as a DataFrame and perform multiclass classification usingOneVsRest. The test error is calculated to measure the algorithm accuracy.

[jkbradley]

Merging with master and branch-1.4

[SparkQA]

Test build #33355 has finished for PR 6296 at commit 46c41b1.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • [OneVsRest](http://en.wikipedia.org/wiki/Multiclass_classification#One-vs.-rest) is an example of a machine learning reduction for performing multiclass classification given a base classifier that can perform binary classification efficiently.
  • [Iris dataset](http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/multiclass/iris.scale), parse it as a DataFrame and perform multiclass classification usingOneVsRest. The test error is calculated to measure the algorithm accuracy.

[SparkQA]

Test build #33356 has finished for PR 6296 at commit 645427c.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • [OneVsRest](http://en.wikipedia.org/wiki/Multiclass_classification#One-vs.-rest) is an example of a machine learning reduction for performing multiclass classification given a base classifier that can perform binary classification efficiently.
  • [Iris dataset](http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/multiclass/iris.scale), parse it as a DataFrame and perform multiclass classification usingOneVsRest. The test error is calculated to measure the algorithm accuracy.