How can we get an association rule object from a flatten collection of association rules objects?
Thank you!!
Alina
0
Answers
Options
homburgModerator, Employee, MemberPosts: 114 RM Data Scientist
Hi Alina,
if you have a collection you can use the "Select" operator to get a single item from it. But I suppose you are talking about the association output of "Create Association Rules". Unfortunately you cannot extract a single rule from it. An option might be found in the handling of the frequent item sets used for those rules. What exactly do you want to do?
Since I have a lot of data I wanted to treat it separatly ie. create association rules on batches of data
Therefore I'm having a collection of association rules but I need to put all the association rules into one single asociaiton rule object in order to be able to apply the rules .
The same thing applyes for selecting one association rule object...
Thank you for your answer!
Alina
0
Options
homburgModerator, Employee, MemberPosts: 114 RM Data Scientist
Hi Alina,
the collection of association rules is a model package that cannot be changed afterwards. The model creation can be controlled by setting the operators parameters or optimizing the frequent item set input. In most cases it is not a good idea to merge different association rule sets since a reasonable rule calculation cannot easily be partitioned. Therefor RapidMiner Studio does not offer a merge operator for rule models. I would suggest to build a representative sample of your data and use it for training a single model. Does such a solution work for your purpose?
Since then I tried Knime and on knime I manage to use all the data (it manages to do it in less than one hour)... that is why I mostly searched for doing the same with RapidMiner
However, I'm havin a lot of struggling with some technical problems (see my RapidAnalytics is slower than RapidMiner and the !process error ...)
Answers
if you have a collection you can use the "Select" operator to get a single item from it. But I suppose you are talking about the association output of "Create Association Rules". Unfortunately you cannot extract a single rule from it. An option might be found in the handling of the frequent item sets used for those rules. What exactly do you want to do?
Cheers,
Helge
Yes you understood it right!
Since I have a lot of data I wanted to treat it separatly ie. create association rules on batches of data
Therefore I'm having a collection of association rules but I need to put all the association rules into one single asociaiton rule object in order to be able to apply the rules .
The same thing applyes for selecting one association rule object...
Thank you for your answer!
Alina
the collection of association rules is a model package that cannot be changed afterwards. The model creation can be controlled by setting the operators parameters or optimizing the frequent item set input. In most cases it is not a good idea to merge different association rule sets since a reasonable rule calculation cannot easily be partitioned. Therefor RapidMiner Studio does not offer a merge operator for rule models. I would suggest to build a representative sample of your data and use it for training a single model. Does such a solution work for your purpose?
Cheers, Helge
Since then I tried Knime and on knime I manage to use all the data (it manages to do it in less than one hour)... that is why I mostly searched for doing the same with RapidMiner
However, I'm havin a lot of struggling with some technical problems (see my RapidAnalytics is slower than RapidMiner and the !process error ...)
Thank you,
Alina