You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@flink.apache.org by "Chenguang He (JIRA)" <ji...@apache.org> on 2016/04/21 22:31:13 UTC

[jira] [Created] (FLINK-3802) Add Very Fast Reservoir Sampling

Chenguang He created FLINK-3802:
-----------------------------------

             Summary: Add Very Fast Reservoir Sampling
                 Key: FLINK-3802
                 URL: https://issues.apache.org/jira/browse/FLINK-3802
             Project: Flink
          Issue Type: Improvement
          Components: Java API
            Reporter: Chenguang He
            Assignee: Chenguang He


Adding Very Fast Reservoir Sampling (http://erikerlandson.github.io/blog/2015/11/20/very-fast-reservoir-sampling/)

An improvement version of Reservoir Sampling, it's used to deal with small sampling in large dataset, where the set of dataset is much larger than the size of sampling.

It is a random sampling proved in the link. The average possibility is P(R/J), where R is size of sampling and J is index of streaming data 

Thanks Erik Erlandson who is the author of this algorithm help me with implementation.



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