Wrapper Method
Last Updated: July 29, 2026 | By Mihail Sebastian | AI Dictionary
A feature selection approach that scores candidate feature subsets by training the model on each one and keeping the subset that performs best.
What is a Wrapper Method?
A wrapper method is a feature selection technique that evaluates feature subsets by training the actual model on each candidate and measuring its performance. The model “wraps around” the search: whichever subset scores best on held-out data wins.
It is one of the three standard approaches to feature selection, alongside filter methods (which score features statistically, without training a model) and embedded methods (which select features during training itself).
How Wrapper Methods Work
A wrapper method needs three parts: a search strategy that proposes subsets, a model to train on each proposal, and a performance metric to compare them. The search continues until adding or removing features stops improving the score.
The common search strategies:
- Forward selection starts with no features and adds the one that improves performance most at each step.
- Backward elimination starts with all features and removes the least useful one at each step.
- Genetic algorithms evolve a population of feature subsets, combining and mutating the strongest candidates.
The trade-off against filter methods is cost versus fidelity. A wrapper trains the model dozens or hundreds of times, which is expensive – but the score reflects how the features perform in that exact model, interactions included, rather than in isolation.
Example of a Wrapper Method
A telecom team predicts customer churn from 40 available features. Forward selection starts with an empty set and tries each feature alone; “months since last contract renewal” gives the best single-feature accuracy, so it stays.
The search repeats. With renewal fixed, “customer service calls” adds the most accuracy; then “monthly charges.” By the fourth round, no remaining feature improves the validation score, and the search stops.
The final model uses 3 features instead of 40. It trains faster, and cross-validation shows less overfitting than the full-featured version.
Related AI terms: Feature Selection · Feature Engineering · Information Gain · Cross-Validation · Overfitting
Did you like the Wrapper Method 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