The concept of taking numeric values and grouping them into finite set of categories so you can treat them as discrete labels instead of continuous numbers
Pros:
- Simplify analysis/visualization
- Reduce noise
- Make models easier to interpret
- Handle non-linear relationships Cons:
- Information loss = values collapse into the same bin
- Boundary sensitivity
- Too many bins = noisy, sparse counts
- Too few bins = oversimplified, hides structure