Implementation:Evidentlyai Evidently Legacy Metric Calculation

Knowledge Sources	Evidentlyai_Evidently
Domains	Metrics, Legacy Compatibility
Last Updated	2026-02-14 12:00 GMT

Overview

Provides the LegacyMetricCalculation abstract base class that bridges legacy Evidently metrics into the modern metric calculation framework.

Description

The Legacy Metric Calculation module defines the LegacyMetricCalculation abstract base class, which acts as an adapter between the legacy metric system (evidently.legacy.base_metric.Metric and MetricResult) and the modern metric calculation system (evidently.core.metric_types.MetricCalculation). This allows existing legacy metrics to be used transparently within the new report and metric pipeline.

LegacyMetricCalculation is a generic class parameterized by four type variables:

TResult -- the modern metric result type
TMetric -- the modern metric type
TLegacyResult -- the legacy metric result type (bound to LegacyMetricResult)
TLegacyMetric -- the legacy metric type (bound to LegacyMetric)

Subclasses must implement two abstract methods:

legacy_metric() -- returns the corresponding legacy metric instance
calculate_value() -- converts the legacy result and widgets into the modern metric result format

The calculate method orchestrates the bridge: it calls the context's get_legacy_metric to execute the legacy metric, then delegates to calculate_value for conversion. It handles both single and paired (current/reference) results and applies default rendering widgets. Additional widgets can be provided via the overridable get_additional_widgets method.

The _gen_input_data method generates InputData from the current Context for the legacy metric execution.

Usage

Use this base class when you need to wrap an existing legacy Evidently metric for use in the modern metric system. Subclass LegacyMetricCalculation, implement legacy_metric and calculate_value, and optionally override get_additional_widgets or _gen_input_data for custom input data generation.

Code Reference

Source Location

Repository: Evidentlyai_Evidently
File: src/evidently/metrics/_legacy.py

Signature

class LegacyMetricCalculation(
    MetricCalculation[TResult, TMetric],
    Generic[TResult, TMetric, TLegacyResult, TLegacyMetric],
    abc.ABC,
):
    @abc.abstractmethod
    def legacy_metric(self) -> TLegacyMetric:
        ...

    def task_name(self) -> Optional[str]:
        ...

    def calculate(self, context: "Context", current_data: Dataset, reference_data: Optional[Dataset]) -> TMetricResult:
        ...

    def get_additional_widgets(self, context: "Context") -> List[BaseWidgetInfo]:
        ...

    @abc.abstractmethod
    def calculate_value(
        self,
        context: "Context",
        legacy_result: TLegacyResult,
        render: List[BaseWidgetInfo],
    ) -> TMetricResult:
        ...

    def _gen_input_data(self, context: "Context", task_name: Optional[str]) -> InputData:
        ...

Import

from evidently.metrics._legacy import LegacyMetricCalculation

I/O Contract

Inputs

Name	Type	Required	Description
context	Context	Yes	The report context providing access to data, metrics, and legacy metric execution
current_data	Dataset	Yes	The current dataset to evaluate
reference_data	Optional[Dataset]	No	An optional reference dataset for comparison
legacy_result	TLegacyResult	Yes (for calculate_value)	The result object produced by the legacy metric
render	List[BaseWidgetInfo]	Yes (for calculate_value)	Widget rendering information from the legacy metric

Outputs

Name	Type	Description
calculate return	TMetricResult	A tuple of (current_result, reference_result) or a single result; the modern metric result with widgets attached
calculate_value return	TMetricResult	The converted modern metric result from legacy metric output
legacy_metric return	TLegacyMetric	The legacy metric instance to be executed
get_additional_widgets return	List[BaseWidgetInfo]	Additional rendering widgets to append (default: empty list)

Usage Examples

from evidently.metrics._legacy import LegacyMetricCalculation
from evidently.core.metric_types import SingleValue, SingleValueMetric
from evidently.legacy.base_metric import Metric as LegacyMetric, MetricResult as LegacyMetricResult
from evidently.legacy.model.widget import BaseWidgetInfo
from typing import List

# Example subclass bridging a legacy metric
class MyLegacyCalculation(
    LegacyMetricCalculation[SingleValue, MyMetric, MyLegacyResult, MyLegacyMetric]
):
    def legacy_metric(self) -> MyLegacyMetric:
        return MyLegacyMetric(param=self.metric.param)

    def calculate_value(
        self, context, legacy_result: MyLegacyResult, render: List[BaseWidgetInfo]
    ) -> SingleValue:
        result = SingleValue(value=legacy_result.score, display_name=self.display_name())
        result.widget = render
        return result

    def display_name(self) -> str:
        return "My Custom Metric"

Related Pages

Environment:Evidentlyai_Evidently_Python_Core_Environment

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment