Classifying as unknown if confidence is under a certain threshold

RyujakkRyujakk Member Posts: 17  Maven
edited November 2018 in Help
Hello there!

So here is my current problem: I'm doing a classification task and I would like examples to be marked as unknown ( '?' ) when the model's confidence is under a certain threshold. I tried using the ThresholdFinder operator, but it seems to only accept binominal exampleSets (whereas I have a polynominal one.) I also tried to include my LibSVMLearner in CostBasedThresholdLearner, which seems like the right track. However, I don't really understand how to configure the CostBasedThresholdLearner to suit my needs...
Am I on the right track here? If not, is there an operator I have not found yet which could work?

Thanks in advance for your help!

- R

Edit: I just found the UncertainPredictionsTransformation operator. You guys really though of everything  :)

Answers

  • landland RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 2,525   Unicorn
    Yes, we are using RapidMiner for predicting the needs of our customers :)
Sign In or Register to comment.