Skip to content

Latest commit

 

History

History
52 lines (35 loc) · 1.71 KB

metrics.rst

File metadata and controls

52 lines (35 loc) · 1.71 KB
Raw

Metrics

Implementation of new stochastic metrics. The metrics assume that the predictions are the parameters of a gaussian distribution. These metrics have to be specified with the arguments stochastic_metrics or stochastic_weighted_metrics when calling the method compile in a subclass of :class:`StochasticModel` :

>>> from src.common.metrics import PredictionIntervalCoverageProbability
>>> from src.model.base_uncertainty_models import StochasticModel
>>> issubclass(model, StochasticModel)
True
>>> model.compile(stochastic_metrics=[PredictionIntervalCoverageProbability()])

The metrics :class:`PICP`, :class:`PINAW` and :class:`CWC` suppose predictions given in update_state are different according the argument input_type :

  1. If input_type="gaussian", the predictions need to be the means and the variances of a gaussian distribution defined as :
  • \hat{\mu} = \text{predictions}[ : , \cdots , : , 0]
  • \hat{\sigma}^2 = \text{predictions}[ : , \cdots , : , 1]
  1. If input_type="pi", the predictions need to be the lower and upper bounds of the predictions defined as :
  • \hat{y}_{lower} = \text{predictions}[ : , \cdots , : , 0]
  • \hat{y}_{upper} = \text{predictions}[ : , \cdots , : , 1]

Here is the list of the new metrics :

PICP

.. autoclass:: purestochastic.common.metrics.PredictionIntervalCoverageProbability
    :members:

PINAW

.. autoclass:: purestochastic.common.metrics.PredictionIntervalNormalizedAverageWidth
    :members:

CWC

.. autoclass:: purestochastic.common.metrics.CoverageWidthBasedCriterion
    :members: update_state