MLS-C01 · Question #386
A machine learning (ML) engineer is preparing a dataset for a classification model. The ML engineer notices that some continuous numeric features have a significantly greater value than most other…
The correct answer is A. Normalize the problematic features. When continuous numeric features have significantly larger values than others, they can disproportionately affect the model's training process, especially for models that rely on distance metrics (e.g., neural networks, logistic regression). Normalization scales these features…
Question
A machine learning (ML) engineer is preparing a dataset for a classification model. The ML engineer notices that some continuous numeric features have a significantly greater value than most other features. A business expert explains that the features are independently informative and that the dataset is representative of the target distribution. After training, the model's inferences accuracy is lower than expected. Which preprocessing technique will result in the GREATEST increase of the model's inference accuracy?
Options
- ANormalize the problematic features.
- BBootstrap the problematic features.
- CRemove the problematic features.
- DExtrapolate synthetic features.
How the community answered
(60 responses)- A88% (53)
- B2% (1)
- C3% (2)
- D7% (4)
Explanation
When continuous numeric features have significantly larger values than others, they can disproportionately affect the model's training process, especially for models that rely on distance metrics (e.g., neural networks, logistic regression). Normalization scales these features to a similar range, ensuring that no feature dominates the learning process due to its magnitude. This typically leads to an improvement in model accuracy, especially when the features are independently informative, as indicated in the problem. Bootstrap the problematic features would involve sampling with replacement to create new datasets, but it doesn't address the issue of feature scaling. Remove the problematic features is not appropriate since the business expert confirmed these features are informative and representative. Extrapolate synthetic features would create new features but doesn't solve the core issue of scale differences in the existing features.
Topics
Community Discussion
No community discussion yet for this question.