AWS Certified Machine Learning Engineer Associate · Free Practice Question Medium
Question 23
A data scientist is building a fraud detection model using Amazon SageMaker. The dataset is highly imbalanced, with only 2% of the data representing fraudulent cases. The scientist wants to ensure the model can effectively detect fraud while minimizing false negatives.
Which techniques should the scientist use to handle this issue? (Choose TWO.)
-
A
Use weighted loss functions during training to penalize misclassification of the minority class.
-
B
Use oversampling techniques, such as SMOTE, to balance the dataset.
-
C
Use Amazon SageMaker Random Cut Forest (RCF) algorithm for anomaly detection.
-
D
Train the model on the original dataset and rely on evaluation metrics for adjustments.
-
E
Perform dimensionality reduction to reduce the number of features and make the dataset more manageable.
Reveal correct answers
Correct answers: A, B
A.
Weighted loss functions assign a higher penalty for misclassifying the minority class, addressing the imbalance during model training.
B.
Oversampling methods like SMOTE can help balance the dataset by generating synthetic examples for the minority class, improving the model's ability to detect fraud.
C.
Random Cut Forest is used for anomaly detection but may not be effective for structured fraud detection tasks.
D.
Training on the original dataset without adjustments can lead to a biased model that underperforms on the minority class.
E.
Dimensionality reduction does not directly address class imbalance and may lead to the loss of important features.
Discussion
Think the marked answer is wrong, or have a better explanation? Share it below — comments appear after review.
