Options

Uploading own Association Rules (not creating them in rapid miner)

JetBlueJetBlue Member Posts: 1 Learner I
edited August 2019 in Help

I have an example list with text information which I want to classify according to content (not sentiment).

Now I have a secound list with set rules - like a dictionary - e.g. if word A,B or C accure it's definitalycategory 1. So the list has the two columns category and words that are 100% related to this categrory.

My question is now how can I uplad this information to my rapid miner model? I can create rules from an example set and apply them, but how do I add an excel list as set of rules?

Answers

  • Options
    kypexinkypexin Moderator, RapidMiner Certified Analyst, Member Posts: 291 Unicorn

    Hi @JetBlue

     

    Generally, there's no way to add static ruleset as a part of any predictive model. Possible solutions depend on your real setup and how exactly are you going to use the rules / model.

     

    On a very general level, if you need to use both rules and model prediction in a real-world production system, this is usually done in a sequence, so first you apply static rules which can potentially filter out some examples which trigger the rules, then you just skip these examples from the model prediction. If you want just to apply a model in RM studio, the same principle applies, you should first filter the examples that can potentially trigger any static rule you have, and exclude those from modelling.    

  • Options
    Thomas_OttThomas_Ott RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,761 Unicorn

    @kypexin love your WTF site. 

  • Options
    MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,524 RM Data Scientist

    Hi @JetBlue,

    you would need to use coding for this. probably an easy groovy script can do this.

     

    Best,

    Martin

    - Sr. Director Data Solutions, Altair RapidMiner -
    Dortmund, Germany
  • Options
    Telcontar120Telcontar120 Moderator, RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,635 Unicorn

    @JetBlue you could also create a predictive model and then implement some manual overrides after the fact using Generate Attributes.  @kypexin talked about how to do that in advance, before going into predictive modeling, but it can also be done afterwards.  But there isn't really a way to seamlessly integrate both manual and machine-learning generated rules into the same "model" object.

    Brian T.
    Lindon Ventures 
    Data Science Consulting from Certified RapidMiner Experts
Sign In or Register to comment.