Connect scoringutils To forecasttools

forecasttools/to_scoringutils.py

@@ -0,0 +1,57 @@
+"""
+Takes differently formatted forecast objects
+and converts them into a format ready for R's
+`scoringutils`. A FluSight submission has all
+necessary content already present save for
+the observation (both for the fitting and
+forecast period). An InferenceData object can
+have the observations available
+"""
+
+import arviz as az
+import polars as pl
+
+
+def from_flusight_to_scoringutils(
+    submission: pl.DataFrame, observations: list[float]
+) -> None:
+    """ """
+
+    pass
+
+
+def from_idata_to_scoringutils(
+    save_path: str,
+    idata: az.InferenceData,
+    observations: list[float],
+    idata_samp_param: str,
+    idata_obs_param: str,
+    overwrite: str = False,
+) -> None:
+    """
+    Take a InferenceData object generated
+    from NumPyro posterior and posterior
+    predictive samples along with observations
+    and generate a parquet file with the
+    correct columns for use in `scoringutils`.
+
+    Parameters
+    ----------
+    save_path
+        The save path for the generated
+        parquet file.
+    idata
+        An InferenceData object containing
+        a forecast made using NumPyro.
+    observations
+        An array of observations corresponding
+        to the
+    overwrite
+        Whether to overwrite an existing file
+        of the same name.
+
+    Returns
+    -------
+    None
+        A parquet file.
+    """

notebooks/notebook-references.bib

@@ -0,0 +1,6 @@
+@article{bosse2022evaluating,
+  title={Evaluating Forecasts with scoringutils in R},
+  author={Bosse, Nikos I and Gruson, Hugo and Cori, Anne and van Leeuwen, Edwin and Funk, Sebastian and Abbott, Sam},
+  journal={arXiv preprint arXiv:2205.07090},
+  year={2022}
+}