Feature Engineering

Last Updated: July 29, 2026 | By Mihail Sebastian | AI Dictionary

The practice of creating, transforming, and combining input variables from raw data so a machine learning model can find patterns more easily.

What is Feature Engineering?

Feature engineering is the practice of creating, transforming, and combining input variables (features) from raw data so a machine learning model can find patterns more easily. In many projects, the quality of the features matters more than the choice of algorithm.

Raw data rarely arrives in the form a model learns from best. A timestamp says little on its own; “day of week” and “hours since last purchase,” derived from that same timestamp, say a lot.

Types of Feature Engineering

  1. Feature transformation reshapes existing variables: scaling numbers to a common range, applying a log to skewed values, or converting categories to numbers through categorical encoding.
  2. Feature creation builds new variables by combining existing ones, such as deriving a debt-to-income ratio from two raw financial columns.
  3. Feature selection then decides which of the resulting features the model keeps.

Feature Engineering vs Feature Selection

Feature engineering creates and transforms features; feature selection chooses among features that already exist. The two run in sequence: engineering expands and reshapes the feature set, selection prunes it down to what carries signal.

CriterionFeature engineeringFeature selection
What it doesCreates or transforms featuresKeeps a subset of existing features
InputRaw or messy dataA finished feature set
OutputNew or modified columnsFewer columns
Typical failureLeaking target information into a featureDiscarding a feature that carried signal

Example of Feature Engineering

A bank builds a credit scoring model. The raw data holds monthly income, outstanding loans, and a log of past payments.

None of these columns predicts default well on its own. So the team engineers new features: debt-to-income ratio (loans divided by income), number of late payments in the past twelve months, and time since the most recent delinquency.

A model trained on these engineered features separates reliable borrowers from risky ones far better than one trained on the raw columns. The information was in the data all along; feature engineering put it in a form the model could use.

Related AI terms: Feature Selection · Categorical Encoding · One-Hot Encoding · Data Leakage · Predictive Modeling

Did you like the Feature Engineering gist?

Learn about 250+ need-to-know artificial intelligence terms in the AI Dictionary.

Mihail Sebastian — Writes about AI governance, regulation, and the technology behind them. Placeholder bio — replace with a real credential line. About

Read the Governor's Letter

Stay ahead with Governor's Letter, the newsletter delivering expert insights, AI updates, and curated knowledge directly to your inbox.

By subscribing to the Governor's Letter, you consent to receive emails from AI Guv.
We respect your privacy - read our Privacy Policy to learn how we protect your information.

Browse All AI Terms A–Z

Every term in the dictionary, in alphabetical order. Jump to a letter or scroll the full list.

A

B

C

D

E

F

G

H

I

J

K

L

M

N

O

P

Q

R

S

T

U

V

W

X

Y

Z