Implementation:Kserve Kserve AnchorTabular Explainer

Knowledge Sources	Kserve_Kserve
Domains	Explainability, Tabular Data
Last Updated	2026-02-13 00:00 GMT

Overview

Concrete tool for generating anchor-based explanations for tabular/structured data models using the Alibi library provided by the KServe sample code.

Description

AnchorTabular extends ExplainerWrapper and wraps the Alibi AnchorTabular explainer. During construction, it receives a prediction function and a pre-built Alibi AnchorTabular explainer instance. The explain() method converts input data to a NumPy array, checks whether the predictor returns class labels or probabilities (applying ArgmaxTransformer if probabilities are returned), updates both the explainer's main predictor and its internal sampler predictor, and then calls the Alibi anchor tabular explanation algorithm on the first instance in the batch. The result produces human-readable rules (e.g., "age > 25 AND income > 50K") that explain individual predictions.

Usage

Use this class as part of the Alibi explainer server pipeline when you need to generate rule-based explanations for structured/tabular data predictions such as credit scoring or classification tasks.

Code Reference

Source Location

• Repository: Kserve_Kserve
• File: docs/samples/explanation/alibi/alibiexplainer/alibiexplainer/anchor_tabular.py
• Lines: 1-58

Signature

class AnchorTabular(ExplainerWrapper):
    def __init__(
        self,
        predict_fn: Callable,
        explainer=Optional[alibi.explainers.AnchorTabular],
        **kwargs,
    ):
        ...

    def explain(self, inputs: List, headers: Dict[str, str] = None) -> Explanation:
        ...

Import

from alibiexplainer.anchor_tabular import AnchorTabular

I/O Contract

Inputs

Constructor

Name	Type	Required	Description
predict_fn	Callable	Yes	Prediction function that accepts input arrays and returns predictions
explainer	Optional[alibi.explainers.AnchorTabular]	Yes	A pre-built Alibi AnchorTabular explainer instance (must not be None)
**kwargs	dict	No	Additional keyword arguments passed to the Alibi explain call

explain()

Name	Type	Required	Description
inputs	List	Yes	List of tabular data rows to explain (first row is used)
headers	Dict[str, str]	No	Optional HTTP headers

Outputs

explain()

Name	Type	Description
anchor_exp	Explanation	Alibi Explanation object containing anchor rules and metadata

Usage Examples

Basic Usage

from alibiexplainer.anchor_tabular import AnchorTabular
import alibi
import numpy as np

# Assume a pre-built AnchorTabular explainer and a predict function
def my_predict_fn(data):
    return np.array([0, 1])

# Build the explainer with training data
anchor_tab_explainer = alibi.explainers.AnchorTabular(
    predictor=my_predict_fn,
    feature_names=["age", "income", "score"]
)
anchor_tab_explainer.fit(training_data)

wrapper = AnchorTabular(predict_fn=my_predict_fn, explainer=anchor_tab_explainer)

# Generate an explanation for a tabular instance
sample_row = [[25, 50000, 0.8]]
explanation = wrapper.explain(sample_row)
print(explanation.anchor)  # e.g., ["age > 20", "income > 40000"]

Related Pages

• Environment:Kserve_Kserve_Python_Runtime