"Generate Products" operator should generate only unique combinations of the attributes

yzanyzan MemberPosts:66Unicorn
edited December 2018 in箴duct Feedback - Resolved

Currently, when we pass a set {"att1", "att2", "att3"} as "attribute1" and {"att1", "att2", "att3"} as "attribute2" to "Generate Products" operator, we get a product between "att1" and "att2" twice. Once as "(att1) * (att2)" and once as "(att2) * (att1)". Since multiplication (for all numerical data types that RapidMiner supports) is commutative, it creates an annecessary redundancy.
箴posal: Generate only unique combinations of the attributes.
Workaround: Use "Remove Correlated Attributes" after "Generate Products".

Tagged:
0
0 votes

Declined·Last Updated

Closing this idea - zero votes since Aug 2018. Please re-open if this is of interest.PROD-834

Comments

  • sgenzersgenzer Administrator, Moderator, Employee, RapidMiner Certified Analyst, Community Manager, Member, University Professor, PM ModeratorPosts:2,959Community Manager

    Wow...been using RapidMiner Studio for six years and never saw this operator before.:)当然乐意开放投票。

  • rfuentealbarfuentealba Moderator, RapidMiner Certified Analyst, Member, University ProfessorPosts:568Unicorn

    Hello@yzan,

    Might I suggest a slight change? What you wrote here makes total sense, and I believe it should be the default: it shouldn't be expected to generate a cross join.

    However, I found myself in more cases where I don't have to remove correlated attributes than the other way round, hence I think the way the product generation is executed should be configurable. (Besides, it reduces the impact in current models).

Sign InorRegisterto comment.