Feature importance is a quantitative measure used in statistical modeling and machine learning to rank input features by their contribution to a model's predictive output. It assigns each feature a score — derived from methods such as permutation importance, SHAP values, coefficient magnitude, or tree-based split gains — that reflects how much the model performance degrades when that feature's information is removed or altered. The parameters are: (1) a trained model, (2) a dataset for evaluation, and (3) a scoring function that maps feature ablation to output change. It persists through model documentation, training logs, and interpretability reports, enabling humans to understand which variables drive predictions and to audit model behavior. [formal: importantia characteris | substrate: mind | horizon: a moment | explicit: yes | epoch: 0.01]
Accepted ontology entry
feature importance
Feature importance is a quantitative measure used in statistical modeling and machine learning to rank input features by their contribution to a model's predictive output. It assigns each feature a score — derived from methods such as perm…
Definition
Why it is in scope
A measure or ranking that quantifies how much each input feature contributes to a model's predictive performance — a construct built by data scientists to interpret and explain machine learning model behavior.
Names and aliases
- feature importanceen · CANONICAL
Relations from this entry
- cmrwiv1rn00a8soacg5vdpiogINSTANCE_OF →
TESTED INSTANCE_OF: feature importance IS a specific kind of metric — a quantitative score measuring a feature's contribution to model output. A competent speaker would call feature importance 'a metric' for model interpretability.
- cmsbl5t8p01du3vv3gxxolk2ySERVES →
TESTED SERVES: feature importance is built and maintained for the sake of model interpretability — its designed purpose is to reveal which features drive model predictions, making black-box models interpretable to humans.
- cmse99x1a04zf3vv3lr34v2hyINSTANCE_OF →
Tested INSTANCE_OF: feature importance IS a specific kind of feature engineering. Feature engineering encompasses the full pipeline of creating, transforming, ranking, and selecting features. Feature importance scoring (computing quantitative measures of each feature's contribution to model performance) is a specific technique within that pipeline. Pinned sense: feature importance as the scoring/ranking subtask of the engineering process.
Relations to this entry
- cmsede9no057x3vv3x9h8b4zr← INSTANCE_OF
Permutation importance IS a specific kind of feature importance — it is a method for ranking features by their contribution to model performance, determined by shuffling each feature's values. A competent speaker would call it a type of feature importance method.
- cmsee23cc059j3vv3e4yxyrdo← INSTANCE_OF
SHAP values IS a specific kind of feature importance: they attribute feature contributions to model predictions using Shapley values from cooperative game theory. A competent speaker would describe SHAP values as a feature importance method. Nearest kind: feature importance.
- cmsk6wtjs03zinobp3vwy6u3r← SERVES
SHAP plots are designed and maintained for the sake of feature importance: their purpose is to visualize, communicate, and make interpretable the importance scores of input features. For whose sake? Feature importance — the plot is the servant, the concept is the master. Law 8d: the servant points at the master.
Record identity
- Created
- Aug 4, 2026, 7:19 AM UTC
- Content hash
- 4159250a36f124797a77b9fd15a101bae8aa8b6b23578bc71d0ab383df9a091a