Implementation:Evidentlyai Evidently Metric Container

Knowledge Sources	Evidentlyai_Evidently
Domains	Metrics, Presets, Data Monitoring
Last Updated	2026-02-14 12:00 GMT

Overview

Abstract base classes for metric containers that programmatically generate multiple related metrics, enabling presets and column-level metric generators.

Description

This module defines the core abstractions for grouping and generating metrics in Evidently:

MetricContainer is an abstract base class (extending EvidentlyBaseModel and AutoAliasMixin) that serves as the foundation for creating collections of metrics. Concrete subclasses implement generate_metrics(context) to produce a sequence of Metric or nested MetricContainer objects. Key features include:

• metrics(context) -- Returns generated metrics with caching via the context's fingerprint system
• render(context, child_widgets) -- Combines visualization widgets from child metrics into a flat list
• list_metrics(context) -- Recursively yields all leaf Metric objects, flattening nested containers
• _get_tests(tests) -- Handles the include_tests flag, converting test specifications or returning defaults
• include_tests -- Boolean flag controlling whether generated metrics include default tests

ColumnMetricContainer is a specialized subclass that adds a column attribute, simplifying containers that generate metrics for a single named column.

MetricOrContainer is a type alias (Union[Metric, MetricContainer]) used throughout the codebase to represent items that are either individual metrics or containers that expand into multiple metrics.

Usage

Subclass MetricContainer to create presets (e.g., DataDriftPreset) or generators (e.g., ColumnMetricGenerator) that produce multiple metrics based on dataset context. Subclass ColumnMetricContainer when the container logic is scoped to a single column.

Code Reference

Source Location

• Repository: Evidentlyai_Evidently
• File: src/evidently/core/container.py

Signature

MetricOrContainer = Union[Metric, "MetricContainer"]

class MetricContainer(AutoAliasMixin, EvidentlyBaseModel, abc.ABC):
    include_tests: bool = True

    def __init__(self, include_tests: bool = True, **data):

    @abc.abstractmethod
    def generate_metrics(self, context: "Context") -> Sequence[MetricOrContainer]:

    def metrics(self, context: "Context") -> List[MetricOrContainer]:

    def render(
        self,
        context: "Context",
        child_widgets: Optional[List[Tuple[Optional[MetricId], List[BaseWidgetInfo]]]] = None,
    ) -> List[BaseWidgetInfo]:

    def list_metrics(self, context: "Context") -> Generator[Metric, None, None]:

    def _get_tests(self, tests):

class ColumnMetricContainer(MetricContainer, abc.ABC):
    column: str

    def __init__(self, column: str, include_tests: bool = True):

Import

from evidently.core.container import MetricContainer, ColumnMetricContainer, MetricOrContainer

I/O Contract

Inputs

Name	Type	Required	Description
include_tests	bool	No (default: True)	Whether generated metrics should include default tests
context	Context	Yes (method arg)	Report context containing datasets and configuration, used for metric generation and caching
column	str	Yes (ColumnMetricContainer)	Name of the target column for column-scoped containers

Outputs

Name	Type	Description
generate_metrics return	Sequence[MetricOrContainer]	Sequence of Metric or MetricContainer objects produced by the container
metrics return	List[MetricOrContainer]	Cached list of generated metrics
list_metrics return	Generator[Metric]	Iterator over all leaf Metric objects (flattened)
render return	List[BaseWidgetInfo]	Combined visualization widgets from child metrics

Usage Examples

from evidently.core.container import MetricContainer, MetricOrContainer
from evidently.core.report import Context
from typing import Sequence

class MyPreset(MetricContainer):
    def generate_metrics(self, context: Context) -> Sequence[MetricOrContainer]:
        # Return a list of metrics for this preset
        return [
            SomeMetric(column="feature_1"),
            SomeMetric(column="feature_2"),
        ]

# Use in a Report
from evidently import Report
report = Report(metrics=[MyPreset()])

Related Pages

• Environment:Evidentlyai_Evidently_Python_Core_Environment
• Implementation:Evidentlyai_Evidently_Core_Tests - Test types used with the include_tests flag
• Implementation:Evidentlyai_Evidently_CLI_Report - CLI command that uses MetricOrContainer in report configuration