MLA-C01 · Question #117
A machine learning engineer is preparing a data frame for a supervised learning task with the Amazon SageMaker Linear Learner algorithm. The ML engineer notices the target label classes are highly…
The correct answer is D. For each feature, approximate the missing values using supervised learning based on other. Option D is correct because using supervised learning to approximate missing values (e.g., regression or KNN imputation) leverages the correlations between features, producing estimates that reflect the actual data structure rather than ignoring it - this is the most…
Question
A machine learning engineer is preparing a data frame for a supervised learning task with the Amazon SageMaker Linear Learner algorithm. The ML engineer notices the target label classes are highly imbalanced and multiple feature columns contain missing values. The proportion of missing values across the entire data frame is less than 5%. What should the ML engineer do to minimize bias due to missing values?
Options
- AReplace each missing value by the mean or median across non-missing values in same row.
- BDelete observations that contain missing values because these represent less than 5% of the
- CReplace each missing value by the mean or median across non-missing values in the same
- DFor each feature, approximate the missing values using supervised learning based on other
How the community answered
(24 responses)- A8% (2)
- B17% (4)
- C4% (1)
- D71% (17)
Explanation
Option D is correct because using supervised learning to approximate missing values (e.g., regression or KNN imputation) leverages the correlations between features, producing estimates that reflect the actual data structure rather than ignoring it - this is the most bias-minimizing approach when feature relationships exist.
Why the distractors fail:
- A (mean/median across the same row) is statistically meaningless - features have different scales, units, and distributions, so averaging across columns produces nonsensical values.
- B (delete rows with missing values) risks introducing selection bias, particularly dangerous here because of class imbalance - missing values are often not random, and deletion disproportionately affects minority-class samples, worsening the imbalance.
- C (mean/median across the same column) is simple imputation - it ignores relationships between features, artificially compresses variance, and is especially problematic with imbalanced targets because it blends statistics across classes.
Memory tip: Think "fight complexity with complexity." When a dataset has class imbalance and correlated features, a naive fix (deletion or simple averages) propagates or worsens existing bias. Supervised imputation uses the data's own structure to fill gaps - it's the only option that treats missing values as an inference problem rather than an inconvenience.
Topics
Community Discussion
No community discussion yet for this question.