[jira] [Commented] (MADLIB-1031) Add parameter to limit itemset size in Apriori

["Nandish Jayaram (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MADLIB-1031?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15830426#comment-15830426 ] 

Nandish Jayaram commented on MADLIB-1031:
-----------------------------------------

In some cases, we do find improvement in the execution time using this parameter.
For instance, running assoc_rules over a particular dataset, with certain support and threshold values resulted in the following:
* max_itemset_size = NULL (max frequent itemset size found = 11)
total_rules: 42295
time: 99.3 seconds

* max_itemset_size = 7
total_rules: 34296
time: 89.1 seconds

* max_itemset_size = 5
total_rules: 13886
time: 43.3 seconds

> Add parameter to limit itemset size in Apriori
> ----------------------------------------------
>
>                 Key: MADLIB-1031
>                 URL: https://issues.apache.org/jira/browse/MADLIB-1031
>             Project: Apache MADlib
>          Issue Type: Improvement
>          Components: Module: Association Rules
>            Reporter: Frank McQuillan
>            Assignee: Nandish Jayaram
>            Priority: Minor
>             Fix For: v1.10
>
>
> Limit itemset size:  see maxlen parameter on p. 10 of 
> https://cran.r-project.org/web/packages/arules/arules.pdf
> This can also serve to improve performance by limiting the number of items.
> This parameter will not change support or confidence computations.



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