I am at the point of reporting my results in a research article conducted. The dataset is highly imbalanced with class 1 and class 4 the minority classes. I am kind of loss how bet to interpret my result in the results discussion section. I computed these metrics as in the table below:
Often I see people compute roc_auc per class, but I would like to know if computing roc_auc_score as a weighted average (as in my case here suffice to give good interpretation of model performance, i.e. using:
skelearn.meretics.roc_auc_score(y_true, y_pred, average='weighted',
multi_class='ovr', labels=[0,1,2,3,4])
