[SOLVED] Applying Association Rule after generating Association Rule

usct01usct01 Member Posts: 10 Contributor II
edited November 2019 in Help
Hi team
Please help me in using process "Applying Association Rules" to a dataset.
I have a data with CustomerID, ProductName, Revenue
I created a set of association rule based on the above metrics (revenue used as weight)
Now I want to create a data set which has one more column(s) showing Recommended product based on association rules created above.

I tried using process"Applying Association Rules", but it asks for some binary variable, actually it says revenue should be binary!!!

Please help and suggest solution to this

Thanks
Deepak

Answers

  • MariusHelfMariusHelf RapidMiner Certified Expert, Member Posts: 1,869 Unicorn
    Hi Deepak,

    most probably you have applied FP-Growth before Create Association Rules. You should pass the same data set you passed into FP-Growth into Create Association Rules.
    If my assumption is not true, please provide your processes and a detailed description of your data.

    Best regards,
    Marius
  • usct01usct01 Member Posts: 10 Contributor II
    Hi Marius,
    Thanks for the reply, It's working now :)

    Now I am struggling to interpret the results of the output after applying Association rules, the results are as below
    List of ID's and columns representing existing product names with values as true/ false (if product is subscribed/not subscribed)
    and few new columns as "Confidence(prod1)", "Confidence(prod2)" etc. having values 1/0.
    I assume that these column may be recommended products, but a close inspection shows some products in recommended list are already subscribed

    Any thoughts on this
    Regards
    Deepak
  • usct01usct01 Member Posts: 10 Contributor II
    Hi All
    Please help me on this, I am stuck!!

    Thanks
    Deepak
  • MariusHelfMariusHelf RapidMiner Certified Expert, Member Posts: 1,869 Unicorn
    usct01 wrote:

    I assume that these column may be recommended products, but a close inspection shows some products in recommended list are already subscribed
    Indeed the 0/1 indicates recommendations. In fact some of the products can already be subscribed. The reasoning is the following. Suppose the following rule:
    [A,C] => B

    And the user has the products [A, B, C]. Then the rule matches, and B is suggested.
    I will add an internal feature request to check for these duplicates, though.

    Best regards,
    Marius
  • usct01usct01 Member Posts: 10 Contributor II
    Ohhkk... Thanks Marius!
Sign In or Register to comment.