Standard Scaling Method

What is Standard Scaling?

Standard Scaling (also called Z-score normalization) is a data preprocessing technique that transforms features to have a mean of 0 and a standard deviation of 1.

Formula

z = (x - mean) / std

Where:
- z: Scaled value
- x: Original value
- mean: Mean of the feature
- std: Standard deviation of the feature

When to Use It

• Algorithms sensitive to scale: SVM, KNN, neural networks
• Gradient-based algorithms: Logistic regression, neural networks
• Distance-based methods: K-means clustering

When NOT to Use It

• Tree-based algorithms: Random Forest, XGBoost (scale-invariant)
• When interpretability matters: Scaled values are harder to interpret

Related Methods

• MinMax Scaling: Scales to a fixed range (0-1)
• Robust Scaling: Uses median and quartiles (handles outliers)

Related Terms

• Random Forest: Scale-invariant algorithm