Glossary term

Feature engineering

Choosing or creating input features that help a model learn better.

How it works

Feature engineering means turning raw data into inputs a model can use well: combining columns, extracting the day of the week from a date, scaling numbers or encoding categories.

It is central to classic machine learning on tables, and less necessary for deep models that learn features themselves.

Feature engineering

Step 1Raw data
Step 2Clean and fill gaps
Step 3Create new columns
Step 4Scale / encode
Step 5Model input

In code

Create useful features from a date

from datetime import date

d = date(2026, 10, 10)
features = {
    "weekday": d.weekday(),          # 0 = Monday
    "is_weekend": d.weekday() >= 5,
    "month": d.month,
}

Why it matters

On tabular business data, thoughtful features often beat a fancier algorithm.

Feature engineering

How it works

In code

Why it matters

Related terms

Feature engineering

How it works

In code

Why it matters

Related terms