"Association rules"

bkruger

I love Rapidminer more and more every day I work with it!  I have a problem and will appreciate any help please?

I have the following data in a database. The customer ID, Transaction ID and items they bought. I read the data into Rapidminer with the read database operator. I prepare the data, do an FP-Growth and Association rules – all working good and I get good rules.

Customer  Transaction  IItem1  Item2  Item3  Item4
1              1                    A          B
1              2                    A          B        C
1              3                    A                    C        D
1              4                                B        C        D
2              1                    A                    C
2              2                                            C        D
2              3                    A                                D
2              4                                B        C        D

Now, my question:

I am not a Rapidminer expert and I want to know if there is a way to get the frequent item sets PER customer so that I can compare them against each other. Currently it looks at rules across the entire set.

What I do to circumvent this issue is that I add a where clause in my query (WHERE Customer = ‘1’) and then manually take the results to Excel and repeat it for each customer. This is a frustrating slow and “dumb” way – I know. Any ideas on how I can make life easier?

My second question is: How do I get either the FP results or the Association rules results back into the database or CSV for further processing in the DB side? It does not work when you try and cast these results into a WriteDatabase.

Thanks in advance for your ingenious suggestions 

BK

bkruger

Hi again,

No ideas? There must be a way to do a process for a subset of data and then loop through it for each subset?

Cheers
BK

nandadorea

I am trying to figure out the same thing - how to save the frequent item sets. I can see the results if I connect the "fre" output to the results, but I don't want to "see" them, I want to be able to compare them for two subsets.

As for getting the frequent sets for each data subset, I also did it taking the longer route: I split the dataset into two, and ran the FP-Growth node independently for each. It works fine, and I get results for both, but now I need some way of saving these results, so that I can manipulate/compare.

In summary: from the "fre" output port from the FP-Growth node, where can I go? where can I save it to?

nandadorea

If only I'd begin from the start... I fund my answer on the RapidMiner tutorial, and I achieve what I intended using the nodes "create association rules" and "apply association rules", in sequence with the FP-Growth. the output of the last one is back as example dataset. Hope you can have some luck as well.