Normalization vs Standardization: Why They’re Not the Same

People treat these two as interchangeable. they’re not.

👉 Normalization (Min-Max scaling):
Compresses values to 0–1.
Useful when magnitude matters (pixel values, distances).

👉 Standardization (Z-score):
Centers data around mean=0, std=1.
Useful when distribution shape matters (linear/logistic regression, PCA).

🔑 Key idea:
Normalization preserves relative proportions.
Standardization preserves statistical structure.

Pick the wrong one, and your model’s geometry becomes distorted.