lenskit.metrics.Metric
======================

.. py:class:: lenskit.metrics.Metric[L, S: MetricResult]
   :canonical: lenskit.metrics._base.Metric

   Bases: :py:obj:`abc.ABC`, :py:obj:`lenskit.data.accum.AccumulatorFactory`\ [\ :py:obj:`L`\ , :py:obj:`S`\ ]


   Base class for LensKit metrics.  Simple metrics that compute a single value
   for a list should extend :class:`ListMetric`.

   This class defines the interface for metrics. Subclasses should implement
   the `measure_list` method to compute metric values.

   The `summarize()` method has a default implementation that computes the mean
   of the per-list metric values, but subclasses can override it to provide
   more appropriate summary statistics.

   .. versionchanged:: 2026.1

       Removed the ``summarize`` method in favor of requiring metrics to
       implement :class:`AccumulatorFactory` to allow metric-controlled
       accumulation.

   :Stability: Full

   .. note::

       For simplicity in the analysis code, you cannot simply implement the
       properties of this class on an arbitrary class in order to implement a
       metric with all available behavior such as labeling and defaults; you
       must actually extend this class. This requirement may be relaxed in the
       future.

   The default value to impute when computing statistics over missing values.
   If ``None``, no imputation is done (necessary for metrics like RMSE, where
   the missing value is theoretically infinite).


   .. py:property:: label
      :type: str


      The metric's default label in output.
      The base implementation returns the class name by default.


   .. py:method:: __str__()


   .. py:method:: measure_list(output, test, /)
      :abstractmethod:


      Compute measurements for a single list.

      :returns:

                - A float for simple metrics
                - Intermediate data for decomposed metrics
                - A dict mapping metric names to values for multi-metric classes



   .. py:method:: extract_list_metrics(data, /)

      Extract per-list metric(s) from intermediate measurement data.

      :returns:

                - A float for simple metrics
                - A dict mapping metric names to values for multi-metric classes
                - None if no per-list metrics are available



   .. py:method:: create_accumulator()
      :abstractmethod:


      Creaet an accumulator to aggregate per-list measurements into summary
      metrics.

      Each result from :meth:`measure_list` is passed to
      :meth:`Accumulator.add`.