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
- 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.
- Feature creation builds new variables by combining existing ones, such as deriving a debt-to-income ratio from two raw financial columns.
- 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.
| Criterion | Feature engineering | Feature selection |
|---|---|---|
| What it does | Creates or transforms features | Keeps a subset of existing features |
| Input | Raw or messy data | A finished feature set |
| Output | New or modified columns | Fewer columns |
| Typical failure | Leaking target information into a feature | Discarding 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