MLA-C01 Practice Question: A financial services company uses SageMaker to…

A financial services company uses SageMaker to train a fraud detection model. They have imbalanced data with 1% fraud. They trained a Gradient Boosting model using SMOTE for oversampling and achieved 99% accuracy on the test set, but the fraud recall is only 10%. The data scientist is concerned about the model's performance. Which change is most likely to improve fraud recall without sacrificing too much precision?

⚠ Common exam trap

MLA-C01 often tests the misconception that changing the evaluation metric (like switching to F1-score) will improve model performance — it only changes measurement, not the model's learned behavior.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Increase the weight of the fraud class in the loss function.

The model achieves 99% accuracy but only 10% fraud recall because the class imbalance causes the loss function to be dominated by the majority (non-fraud) class. Increasing the weight of the fraud class in the loss function (e.g., via scale_pos_weight in XGBoost or class_weight in scikit-learn) directly penalizes misclassification of fraud cases more heavily, forcing the model to prioritize recall on the minority class. This is a targeted intervention at the learning objective itself, unlike changing the evaluation metric, which only changes how you measure performance without altering what the model learns.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Use a different evaluation metric like F1-score during training.
Why it's wrong here
Switching the training metric to F1-score changes only how models are scored and selected; it does not alter the class distribution or decision threshold that suppress fraud recall. It is tempting because F1 balances precision and recall, and it would be correct where the algorithm already learns from balanced data but model selection ignores the minority class.
✓
Increase the weight of the fraud class in the loss function.
Why this is correct
Class weighting directly penalises misclassified fraud cases during gradient boosting, pushing the model to raise recall on the 1% minority class. Unlike SMOTE, which already oversampled, weighting alters the loss landscape itself, improving fraud detection while retaining most precision.
✗
Reduce the SMOTE sampling ratio to create more synthetic samples.
Why it's wrong here
Reducing the SMOTE ratio generates fewer synthetic fraud examples, moving the training distribution back toward the 1% imbalance that produced the 10% recall. It is tempting because lowering synthetic volume limits overfitting to fabricated minority points, and it would be correct if SMOTE were already oversampling beyond the desired balance.
✗
Use a random undersampling of the majority class.
Why it's wrong here
Random undersampling discards most legitimate transactions, so the model sees far less majority-class signal and precision typically collapses alongside the recall gain. It is tempting because undersampling does rebalance classes quickly, and it would be correct where the majority class is small enough that removing records costs little information.

Go deeper

MLA-C01 Practice Question: A financial services company uses SageMaker to…

Why each option matters

Option-by-option breakdown

Related to this question