1
$\begingroup$

I use to gradient boosting to classify my data between default and paid. The data is very imbalanced where default is in minority. The fisrt classification report from sklearn gradient boosting classifier give:

             precision    recall  f1-score   support

   Defaulted       0.56      0.03      0.05       364
        Paid       0.87      1.00      0.93      2420

   micro avg       0.87      0.87      0.87      2784
   macro avg       0.71      0.51      0.49      2784
weighted avg       0.83      0.87      0.82      2784

and an AUC = 0.614

To addressed the imbalence issue, I use SMOTE to upsample the default class
and I am getting an AUC = 0.941

However, the classification report wost f1-score for the default class:

              precision    recall  f1-score   support

   Defaulted       0.38      0.01      0.02       364
        Paid       0.87      1.00      0.93      2420

   micro avg       0.87      0.87      0.87      2784
   macro avg       0.62      0.50      0.47      2784
weighted avg       0.81      0.87      0.81      2784

Is this possible and how should I interpret or understand this difference between the worst f1-score but a better AUC?

$\endgroup$

Your Answer

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

Browse other questions tagged or ask your own question.