What is feature binning and why is it used in feature engineering?
Updated May 16, 2026
Short answer
Feature binning converts continuous variables into discrete intervals to reduce noise and capture non-linear patterns.
Deep explanation
Feature binning (or discretization) groups continuous numerical values into bins or categories. This helps reduce the effect of noise, handle outliers, and improve model performance for algorithms that benefit from categorical representations. It is especially useful in credit scoring and risk modeling where ranges matter more than exact values.
Real-world example
Credit risk models often bin income into ranges like low, medium, and high instead of using raw salary values.
Common mistakes
- Creating too many bins which leads to overfitting or losing interpretability by using arbitrary bin edges.
Follow-up questions
- What is equal-width vs equal-frequency binning?
- Does binning reduce model accuracy always?