Feature values range from 1 to 1,000,000 and affect a distance-based model. What helps?
Why D is the answer
Scaling or normalizing the features
Scaling stops large-valued features from dominating distance calculations.
In detail
- Scaling stops a large range from dominating distances.
- Fit the scaler on training data only.
Use case
An illustrative scenario, not a real incident.
Income runs to a million and age to 100, so income dominates distances. Scaling puts them on comparable ranges.
See it in different ways
flowchart LR A["Raw data in S3"] --> B["Clean and transform"] B --> C["Feature engineering"] C --> D["Check quality and bias"] D --> E["Training set"]
mindmap
root((Data preparation))
Missing values
Impute
Categories
One-hot encoding
Ranges
Scaling
Imbalance
Resampling
Class weights
Leakage
Remove future dataflowchart LR S0["Raw features"] --> S1["Scale"] --> S2["Train distance-based model"]
Key point: Scale for distance-based models.
Why the other options are wrong
- A. One-hot encoding each numeric value
- One-hot encoding is for categories, not numeric ranges.
- B. Removing the validation set
- The validation set has nothing to do with feature ranges.
- C. Adding a dropout layer
- Dropout is regularization, and it does not change input ranges.