Setting a
abaker
New Altair Community Member
Hi, I'm new to RapidMiner, so perhaps this is a silly question, or I'm missing an obvious detail. My nomenclature may be off as well, hopefully I convey my issue.
For classification problems (let's say we're using Logistic Regression, Decision Tree, or another common approach), is there a way to have RapidMiner understand a specific class to be the "class of interest"? That is, when I run performance operators, I'd like to be sure that my Sensitivity, Specificity, True Positive, etc. rates are all reported with respect to the focal class that I specifically identified.
Is there a specific Operator that allows for this?
Is there a "clever workaround" for dealing with this?
Something else?
Thanks!
For classification problems (let's say we're using Logistic Regression, Decision Tree, or another common approach), is there a way to have RapidMiner understand a specific class to be the "class of interest"? That is, when I run performance operators, I'd like to be sure that my Sensitivity, Specificity, True Positive, etc. rates are all reported with respect to the focal class that I specifically identified.
Is there a specific Operator that allows for this?
Is there a "clever workaround" for dealing with this?
Something else?
Thanks!
Tagged:
0
Best Answer
-
Hello @abaker
Can you please go through the below link? Looks like a similar topic discussed here
https://community.rapidminer.com/discussion/52371/positive-class-assigning-logic
Please inform if this is not the case.
Thank you2
Answers
-
Hello @abaker
Can you please go through the below link? Looks like a similar topic discussed here
https://community.rapidminer.com/discussion/52371/positive-class-assigning-logic
Please inform if this is not the case.
Thank you2 -
Thank you! I was searching using any verbiage I could imagine, but "map" wasn't one I thought of, thank you very much! Easy solution!3
-
Hi @abaker,
9.3 has a new feature to explicitly state the positive class in Performance (Binominal). It is currently in Beta.
Best,
Martin2