Numerical Features
← Back to Feature Engineering
Transforming numeric data for optimal model performance. Many algorithms are sensitive to feature scale and distribution.
Techniques
- Standardization — zero mean, unit variance (z-score)
- Normalization — scale to [0, 1] range (min-max)
- Binning — convert continuous to discrete ranges
- Log Transforms — handle skewed distributions
Related
- Categorical Features (non-numeric transformations)
- K-Nearest Neighbors (scaling critical for distance-based methods)