0
$\begingroup$

I split my dataset into 2 parts: 75% of it is the training set, 25% of it is the test set. Then I estimated the logistic regression parameters in the training set and I compute the Area Under the ROC Curve (AUC) (of the model estimated in the training set) from the test set. Since the test set is formed by 2500 observations on 20 variables, I was expected to get 2500 AUC, one each row. Why I just get one AUC?

$\endgroup$
  • 6
    $\begingroup$ Area under the ROC curve is a summary statistics for the performance on your entire test set. It quantifies your model's performance on all 2500 observations, yielding a single number. $\endgroup$ – Marc Claesen Oct 22 '15 at 12:02
  • $\begingroup$ @MarcClaesen this should really be written as an answer instead of a comment... $\endgroup$ – Calimo Oct 22 '15 at 16:33
  • $\begingroup$ @Calimo done, just feels so ... short :-) $\endgroup$ – Marc Claesen Oct 22 '15 at 16:41
2
$\begingroup$

Area under the ROC curve is a summary statistics for the performance on your entire test set. It quantifies your model's performance on all 2500 observations, yielding a single number.

$\endgroup$
  • $\begingroup$ Thanks for the explanation Marc. So, why I still get one AUC if my test set is formed by just 1 observation on all the variables? $\endgroup$ – Luca Dibo Oct 22 '15 at 17:27
  • $\begingroup$ @LucaDibo you will need at least one positive and one negative observation to compute an AUC. $\endgroup$ – Calimo Oct 28 '15 at 17:26

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.