Confidence and Probability

SA_HSA_H MemberPosts:29Contributor II
edited December 2019 inHelp
Could you please let me know what is the difference (mathematically and statistically) between Confidence and Probability.
Also in case of Decision tree algorithm, how to convert confidence to probability.

Tagged:

Answers

  • MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University ProfessorPosts:3,423RM Data Scientist
    Hi,

    Confidence is just a measure, of how sure the algorithm is. It is always normalized to [0-1]. But just because a decision tree gives you a confidence of 0.9, it does not mean that it happens in 90% of the cases.

    Have a look on Rescale Confidence (Logistic) or search for Platt Scaling. That should help to understand how to transform.

    BR,
    Martin

    - Sr. Director Data Solutions, Altair RapidMiner -
    Dortmund, Germany
    BalazsBarany kypexin
  • SA_HSA_H MemberPosts:29Contributor II
    Thank you Martin for your reply. I am trying to find any paper or manual that can explain how rapidminer estimate the confidence value but I can not find till now. I am confused, since probability also can be in the range between zero and one.

    Please for using re-scale confidence operator, do this operator can work with the output of any machine learning classifiers or limited to a certain classifier.
Sign InorRegisterto comment.