ANCHOR

Description

ANCHOR explains one prediction of a black-box classifier with an if-then rule, called an anchor, such as 'age > 30 AND income < £50k'. An anchor names a small set of feature conditions under which the model keeps giving the same prediction, whatever the other features are. To find one, the method fixes the conditions, varies the other features by sampling, asks the model for a prediction each time, and searches for the broadest rule that holds in at least a chosen share of the samples, commonly 95%. Each anchor comes with its estimated precision and its coverage, the share of cases to which it applies. Both figures describe the model's behaviour on the sampled variations, not on real cases, and continuous features are first grouped into ranges.

Example Use Cases

Explainability

A model-validation team at a lender computes anchors for a sample of declined applications from a gradient-boosted credit model. It checks that the conditions holding each decision, such as a low credit-score band together with recent missed payments, match credit policy, and reads each rule's coverage to see how many other applications it applies to.

A hospital data team explains individual alerts from a tabular sepsis-risk model with anchors, and a clinician checks whether the conditions in each rule, such as raised temperature and white-cell count, are clinically sensible or point to an artefact such as the ward an admission came from.

Transparency

An employer's compliance team gives a reviewer the anchor behind a shortlisting prediction, with its precision and coverage. The reviewer tests the rule on other applications and can challenge it if it leans on an attribute that should not matter, such as age.

Limitations

The stated precision is measured on perturbed samples from a chosen sampling distribution, not on real cases. If that distribution produces unrealistic inputs, or is skewed by an unbalanced training set, the figure can overstate how far the rule can be relied on.
An anchor covers only the inputs its conditions match, often a small share. In the original experiments average coverage ranged from about 1% to 29% on tabular data, so one anchor says little about the model as a whole. Choosing a set of anchors helps but adds effort.
Predictions close to the model's decision boundary, or of rare classes, need long, very specific rules with little coverage, which add little insight. Strict precision thresholds make this more likely.
Continuous features must be grouped into ranges before rules can be stated, and the grouping changes which anchors are found. Too coarse hides boundaries, and too fine gives narrow rules with little coverage.
Each anchor needs many queries to the model, so run time depends on how fast the model is and on the number of features. The method is built for classifiers, so a numeric output must first be turned into classes.

Resources

Research Papers

Anchors: High-Precision Model-Agnostic Explanations

Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin•Jan 1, 2018

We introduce a novel model-agnostic system that explains the behavior of complex models with high-precision rules called anchors, representing local, "sufficient" conditions for predictions. We propose an algorithm to efficiently compute these explanations for any black-box model with high-probability guarantees. We demonstrate the flexibility of anchors by explaining a myriad of different models for different domains and tasks. In a user study, we show that anchors enable users to predict how a model would behave on unseen instances with less effort and higher precision, as compared to existing linear explanations or no explanations.

Software Packages

anchor

Feb 2, 2018

Code for "High-Precision Model-Agnostic Explanations" paper

alibi

Feb 26, 2019

Algorithms for explaining machine learning models

Tutorials

Interpretable Machine Learning - Anchors

Jan 1, 2019

Anchors examples — Alibi Explain documentation

Sep 23, 2025

Documentations

Anchors — Alibi Explain documentation

Jul 8, 2024

Related Techniques

Name	Description	Assurance Goals
RuleFit	RuleFit builds a predictive model that a person can read: a short list of if-then rules plus a few ordinary feature weights. It first trains an ensemble of decision trees (many small trees whose outputs are combined) and turns each branch into a candidate rule, such as 'IF age > 50 AND income < £30k'. It then fits a linear model over those rules and the original features using the lasso, a penalty that sets many weights to exactly zero, so only the most useful terms remain. The surviving rules and weights can be read directly, and each term has an importance score, overall and for single predictions. Fitted to real outcomes, the result can replace an opaque model; fitted to an opaque model's predictions, it is a surrogate that approximates that model, and its fit should then be checked.	Explainability Transparency
Ridge Regression Surrogates	A ridge regression surrogate approximates a complex model with a linear equation fitted to the complex model's own predictions. The complex model is run on a set of inputs, and ridge regression, a linear model with a penalty that keeps coefficients stable when features are correlated, is fitted to reproduce its outputs. Each coefficient then shows roughly how far a feature moves the complex model's output, averaged over those inputs. Fidelity, often reported as R-squared (the share of the model's variation the surrogate reproduces on held-out inputs), shows how far the coefficients can be trusted. They describe the complex model, not the real world.	Explainability Transparency
Local Interpretable Model-Agnostic Explanations	LIME (Local Interpretable Model-agnostic Explanations) explains individual predictions by approximating the complex model's behaviour in a small neighbourhood around a specific instance. It works by creating perturbed versions of the input (e.g., removing words from text, changing pixel values in images, or varying feature values), obtaining the model's predictions for these variations, and training a simple interpretable model (typically linear regression) weighted by proximity to the original instance. The coefficients of this local surrogate model reveal which features most influenced the specific prediction.	Explainability Transparency