When running AutoML on imbalanced data, you should still:

Question

Accepted Answer

Inspect class balance, metrics beyond accuracy, and review generated notebooks critically — AutoML + imbalance: check priors, non-accuracy metrics, read the notebooks.

Answer

Trust accuracy alone blindly

Answer

Avoid looking at confusion matrices

Answer

Disable MLflow logging

When running AutoML on imbalanced data, you should still:

Short Explanation and Infographic

Full Explanation