[ 
https://issues.apache.org/jira/browse/SPARK-20265?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Cyril de Vogelaere updated SPARK-20265:
---------------------------------------
    Description: 
Given the sequence :

0 4 0 1 0 2 0 5 0 2 0 4 0 5

And supposing only item 4 and 5 are frequent. The current implementation will 
clean the sequence as follow.

0 4 0 0 0 5 0 0 4 0 5

Which, while still giving the same results, inneficiently leave sequences of 
zero. A quick change in the code can removes those sequence of zeros, while not 
changing the returned solutions.

I ran test to mesure efficiency gain, they are available in the graph below :
https://postimg.org/image/9x6ireuvn/

As you can see, the performance increase is signifiant.

  was:
Given the sequence :

0 4 0 1 0 2 0 5 0 2 0 4 0 5

And supposing only item 4 and 5 are frequent. The current implementation will 
clean the sequence as follow.

0 4 0 0 0 5 0 0 4 0 5

Which, while still giving the same results, inneficiently leave sequences of 
zero. A quick change in the code can removes those sequence of zeros, while not 
changing the returned solutions.

I ran test to mesure efficiency gain, they are available in the graph below :
http://i67.tinypic.com/29ljswy.jpg

As you can see, the performance increase is signifiant.


> Improve Prefix'span pre-processing efficiency
> ---------------------------------------------
>
>                 Key: SPARK-20265
>                 URL: https://issues.apache.org/jira/browse/SPARK-20265
>             Project: Spark
>          Issue Type: Improvement
>          Components: MLlib
>    Affects Versions: 2.1.0
>            Reporter: Cyril de Vogelaere
>            Priority: Minor
>   Original Estimate: 0h
>  Remaining Estimate: 0h
>
> Given the sequence :
> 0 4 0 1 0 2 0 5 0 2 0 4 0 5
> And supposing only item 4 and 5 are frequent. The current implementation will 
> clean the sequence as follow.
> 0 4 0 0 0 5 0 0 4 0 5
> Which, while still giving the same results, inneficiently leave sequences of 
> zero. A quick change in the code can removes those sequence of zeros, while 
> not changing the returned solutions.
> I ran test to mesure efficiency gain, they are available in the graph below :
> https://postimg.org/image/9x6ireuvn/
> As you can see, the performance increase is signifiant.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

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

Reply via email to