You recently deployed an ML model. Three months after deployment, you notice that your model is underperforming on certain subgroups, thus potentially leading to biased results. You suspect that the…

Question

Options

• ARemove training examples of high-performing subgroups, and retrain the model.
• BAdd an additional objective to penalize the model more for errors made on the minority class, and
• CRemove the features that have the highest correlations with the majority class.
• DUpsample or reweight your existing training data, and retrain the model
• ERedeploy the model, and provide a label explaining the model's behavior to users.

• B
  81% (13)
• C
  13% (2)
• E
  6% (1)

Explanation

When class imbalance causes biased performance across subgroups and more data cannot be collected, two standard techniques apply. D (Upsample/reweight): Oversampling minority-class examples or assigning higher sample weights rebalances the effective class distribution during training, directly addressing imbalance. B (Custom loss penalty): Adding a cost-sensitive objective that penalizes errors on minority-class examples more heavily steers the optimizer to reduce those errors disproportionately. Option A (removing majority examples) is destructive and discards potentially useful information. Option C (removing correlated features) could harm the model's ability to learn rather than fixing the imbalance. Option E (redeploy with a label) is a communication measure, not a modeling fix.

No community discussion yet for this question.