0
$\begingroup$

I developed predictive models and I reported G-mean ,F-measure, Sensitivity, and Specificity which are very low. i wonder if anyone has a reference for the threshold of these measures ? I want to report that model is not good enough however I cant find anything in the literature? Thank yo so much and I really appreciate it in advance. Cheers, Ben

$\endgroup$

migrated from stackoverflow.com Apr 27 '18 at 18:22

This question came from our site for professional and enthusiast programmers.

0
$\begingroup$

have you used confusionMatrix in the caret package?

If you use train and test sets it will report the accuracy and the no-information-rate, the former having to be larger than the former for the model to have any predictive value.

eg:

          Reference
Prediction  T   F
       T   39  24
       F   24  52

               Accuracy : 0.6547          
                 95% CI : (0.5694, 0.7332)
    No Information Rate : 0.5468          
    P-Value [Acc > NIR] : 0.006344  
$\endgroup$
  • $\begingroup$ i used WEKA, and for the research related purposes I have to report G-mean , F-measure , Sn and Sp. However I can't find any threshold for these for so that I can accept or reject the model $\endgroup$ – Benn Apr 24 '18 at 13:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.