Source code for ignite.contrib.metrics.regression.mean_absolute_relative_error

from typing import Tuple

import torch

from ignite.contrib.metrics.regression._base import _BaseRegression
from ignite.exceptions import NotComputableError
from ignite.metrics.metric import reinit__is_reduced, sync_all_reduce


class MeanAbsoluteRelativeError(_BaseRegression):
    r"""Calculate Mean Absolute Relative Error.

    .. math::
        \text{MARE} = \frac{1}{n}\sum_{j=1}^n\frac{\left|A_j-P_j\right|}{\left|A_j\right|}

    where :math:`A_j` is the ground truth and :math:`P_j` is the predicted value.

    More details can be found in the reference `Botchkarev 2018`__.

    - ``update`` must receive output of the form ``(y_pred, y)`` or ``{'y_pred': y_pred, 'y': y}``.
    - `y` and `y_pred` must be of same shape `(N, )` or `(N, 1)`.

    __ https://arxiv.org/ftp/arxiv/papers/1809/1809.03006.pdf

    Parameters are inherited from ``Metric.__init__``.

    Args:
        output_transform: a callable that is used to transform the
            :class:`~ignite.engine.engine.Engine`'s ``process_function``'s output into the
            form expected by the metric. This can be useful if, for example, you have a multi-output model and
            you want to compute the metric with respect to one of the outputs.
            By default, metrics require the output as ``(y_pred, y)`` or ``{'y_pred': y_pred, 'y': y}``.
        device: specifies which device updates are accumulated on. Setting the
            metric's device to be the same as your ``update`` arguments ensures the ``update`` method is
            non-blocking. By default, CPU.

    Examples:
        To use with ``Engine`` and ``process_function``, simply attach the metric instance to the engine.
        The output of the engine's ``process_function`` needs to be in format of
        ``(y_pred, y)`` or ``{'y_pred': y_pred, 'y': y, ...}``.

        .. include:: defaults.rst
            :start-after: :orphan:

        .. testcode::

            metric = MeanAbsoluteRelativeError()
            metric.attach(default_evaluator, 'mare')
            y_true = torch.tensor([1., 2., 3., 4., 5.])
            y_pred = y_true * 0.75
            state = default_evaluator.run([[y_pred, y_true]])
            print(state.metrics['mare'])

        .. testoutput::

            0.25...

    .. versionchanged:: 0.4.5
        - Works with DDP.
    """
    _state_dict_all_req_keys = ("_sum_of_absolute_relative_errors", "_num_samples")

    @reinit__is_reduced
    def reset(self) -> None:
        self._sum_of_absolute_relative_errors = torch.tensor(0.0, device=self._device)
        self._num_samples = 0

    def _update(self, output: Tuple[torch.Tensor, torch.Tensor]) -> None:
        y_pred, y = output[0].detach(), output[1].detach()
        if (y == 0).any():
            raise NotComputableError("The ground truth has 0.")
        absolute_error = torch.abs(y_pred - y.view_as(y_pred)) / torch.abs(y.view_as(y_pred))
        self._sum_of_absolute_relative_errors += torch.sum(absolute_error).to(self._device)
        self._num_samples += y.size()[0]

    @sync_all_reduce("_sum_of_absolute_relative_errors", "_num_samples")
    def compute(self) -> float:
        if self._num_samples == 0:
            raise NotComputableError(
                "MeanAbsoluteRelativeError must have at least one sample before it can be computed."
            )
        return self._sum_of_absolute_relative_errors.item() / self._num_samples