Young machine learning practitioners : That is why you should do not trust accuracy when doing a ML model on imbalanced datasets.

And in general, better use precision, recall, f1-score or confusion matrices