Strange rules - what can I do?

CharlieGCharlieG Member Posts: 9 Contributor II
edited November 2018 in Help
I'm building a rule induction model that's working well but is producing some strange rules.  An example is:

if AvgClausRxrT > 453.805 and AvgClausRxrT > 499.515 then Down  (18 / 3)

Even I can simplify this rule :).  While these rule don't cause mistakes they do lower confidence in the results if a human expert sees them and gets to wondering - "If the program is dumb enough to produce this rule whay should I believe the rest of what it's doing?"

Does anyone have a suggestion on how to configure RapidMiner so it doesn't produce these sorts of flaky rules?  Or is that just the way it is and I should learn to live with it?

Answers

  • MariusHelfMariusHelf RapidMiner Certified Expert, Member Posts: 1,869 Unicorn
    Well, indeed this is strange. If your data is not confidential, I would like to ask you to send me a copy of your data and your process, such that we can reproduce and investigate the issue.

    Anyway, as long as you validate your model with a X-Validation and you are satisfied, I would not worry a lot and for take it just the way it is for now ;)

    Best regards,
    Marius
Sign In or Register to comment.