Glossary term
Glossary term
Foundations
A scaling technique that replaces a raw feature value with a floating-point value representing the number of standard deviations from that feature's mean. For example, consider a feature whose mean is 800 and whose standard deviation is 100. The following table shows how Z-score normalization would map the raw value to its Z-score:
The machine learning model then trains on the Z-scores for that feature instead of on the raw values.
Created for this library
A risk modeling team applies Z-score normalization to continuous features before training so coefficients are comparable across features.
A computer vision team applies Z-score normalization per channel on input images so training is faster and more stable.
A retail analytics team applies Z-score normalization to its tabular features so its linear baseline is not dominated by features with large numeric ranges.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License