spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Egor Pakhomov (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SPARK-3509) Method for generating random LabeledPoints for testing
Date Fri, 12 Sep 2014 13:59:33 GMT

    [ https://issues.apache.org/jira/browse/SPARK-3509?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14131554#comment-14131554
] 

Egor Pakhomov commented on SPARK-3509:
--------------------------------------

So far I have bad code for my usages. need good code. bad code  -
  def randomRegressionLabeledFeatureSet(size: Int, featureNumber: Int) = {
    // bad code. task for better code  - SPARK-3509
    val seed = Random.nextLong();
    sc.parallelize(1 to size, 10).map(i => {
      val features = (1 to featureNumber).map(_ => Random.nextDouble()).toArray
      var seedCopy = seed
      val result = features.reduceLeft((a, b) => {
        if (seedCopy % 3 == 0) {
          seedCopy = seedCopy / 3
          a * b
        } else {
          seedCopy = seedCopy / 2
          a + b
        }
      })
      new LabeledPoint(result, Vectors.dense(features))
    })
  }

> Method for generating random LabeledPoints for testing
> ------------------------------------------------------
>
>                 Key: SPARK-3509
>                 URL: https://issues.apache.org/jira/browse/SPARK-3509
>             Project: Spark
>          Issue Type: New Feature
>          Components: MLlib
>    Affects Versions: 1.2.0
>            Reporter: Egor Pakhomov
>            Priority: Minor
>             Fix For: 1.2.0
>
>
> During testing I need random LabeledPoints with some correletion behind it.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org


Mime
View raw message