Course contentsShow
← Machine Learning and Deep Learning
Lesson 408 of 3,53810. Feature EngineeringFree lesson

Min-Max Normalization

Scaling features to a fixed range [0,1] or [-1,1] by subtracting minimum and dividing by range.

Min-Max Normalization

What You'll Learn

You'll learn how to transform features into a fixed range—typically [0,1] or [-1,1]—using a simple mathematical formula that preserves relative distances.

The Core Idea

Min-Max Normalization (also called min-max scaling) squeezes all your feature values into a specific range by finding the minimum and maximum values, then rescaling everything proportionally between them.

The formula is straightforward:

X_scaled = (X - X_min) / (X_max - X_min)

This gives you values in the [0,1] range. For [-1,1], you multiply by 2 and subtract 1.

How It Works

Imagine you have exam scores ranging from 40 to 100. Min-max normalization:

  1. Finds the minimum (40) and maximum (100)
  2. Subtracts the minimum from each score: a score of 70 becomes 70 - 40 = 30
  3. Divides by the range (100 - 40 = 60): 30 / 60 = 0.5

So a score of 70 becomes 0.5 in the normalized scale, exactly halfway between 0 and 1.

Why This Matters

Remember from Feature Scaling for KNN and Feature Scaling for K-Means: algorithms that use distance calculations need features on similar scales. Min-max normalization ensures that a feature measuring temperature in Celsius (ranging 0–40) doesn't get drowned out by a feature measuring income (ranging 20,000–100,000).

Unlike standardization, min-max normalization guarantees your data fits within exact bounds—crucial when neural networks expect inputs in [0,1] or when you need guaranteed ranges for certain algorithms.

Key Takeaway: Min-max normalization rescales features to a fixed range by subtracting the minimum and dividing by the range, preserving relative distances while ensuring all features contribute equally to distance-based algorithms.