AUC Calculation within Auto Model

Muhammed_Fatih_

Dear Community, 

is there a possibility within the Auto Model function to generate the AUC quality measure by parallely computing selected classification models? I couldn't find the option to select the performance paramters like e.g. the F-Score, the AUC and so on.   



Thank you in advance for your support! 

Best regards! 

lionelderkrikor

Hi @Muhammed_Fatih_,

If your problem is in fact a binary classification problem, then the binary classification performance is automatically used
and the AUC is thus calculated.
You can see the AUC in the results panel (the final screen of Auto-model).

Regards,

Lionel

lionelderkrikor

Hi @Muhammed_Fatih_,

If you can not get the AUC score it is maybe because your problem is NOT a binary classification problem (in a binary classification problem the label has only two possible values - TRUE/FALSE, YES/NO, 1/0 etc.).
In your case, it seems that your problem is a multi-class classification problem : The number of possible values of your label is strictly greater than 2.In this case , in Auto-Model you only have access to the Accuracy and the classification error.

Regards,

Lionel

Muhammed_Fatih_

Hi @lionelderkrikor

thank you for your answer! My problem is in fact a binary classification problem with two target classes that should be predicted. 

Does the AUC score normally exist in the Auto Model or is the AUC only available via the binary performance operator in the pipeline? 

Many thanks and best regards!  

Muhammed_Fatih_

Hi @lionelderkrikor, 

perfect! The AUC appeared in the result layer of Auto Model as you mentioned. It confused me that the user cannot choose it before. Nevertheless it is automatically generated in case of a binary problem.