Closed dplarson closed 4 years ago
Based on other parts of the codebase, I suggest we standardize on obs
and fx
for observations and forecasts. I'm less sure on input ordering, but I would lean towards (observations, forecasts) since all metrics depend on observations, but some depend on additional inputs related to the forecasts (e.g. obs, fx, fx_prob
rather than fx, fx_prob, obs
).
Good catch. I agree with your suggestions. Let’s double check with @alorenzo175
I also agree with your suggestions.
Great! I'll open a PR to standardize the inputs once PR #257 is merged.
While implementing the deterministic and probabilistic forecast metrics, I forgot to ensure that the function inputs following the same conventions regarding names and ordering of inputs. While the conventions within each type (deterministic and probabilistic) are consistent, they are not the same between the two:
y_true
(observations),y_pred
(forecasts)obs
(observations),fx
(forecasts)For consistency, we should revise the metrics code to use a more consistent naming and ordering convention for the inputs.