fixed available_as_ofs, initial_as_of, and current_date PTC options

src/hub_predtimechart/generate_options.py

@@ -19,6 +19,13 @@ def task_text(task_id, task_value):
             return task_value
 
 
+    def get_max_ref_date_or_first_config_ref_date(reference_dates):
+        if len(reference_dates) == 0:
+            return min(hub_config.reference_dates)
+        else:
+            return max(reference_dates)
+
+
     # set `target_variables` and `initial_target_var`. recall that we currently only support one target
     options = {}
     options['target_variables'] = [{'value': hub_config.fetch_target_id,
@@ -39,9 +46,12 @@ def task_text(task_id, task_value):
     options['initial_interval'] = options['intervals'][-1]
 
     # set `available_as_ofs`, `initial_as_of`, and `current_date`
-    options['available_as_ofs'] = {hub_config.fetch_target_id: hub_config.reference_dates}
-    options['initial_as_of'] = hub_config.reference_dates[-1]
-    options['current_date'] = hub_config.reference_dates[-1]
+    # available_as_ofs is the subset of hub_config.reference_dates for which
+    # there is at least one model output file
+    options['available_as_ofs'] = hub_config.get_available_as_ofs()
+    options['initial_as_of'] = max([get_max_ref_date_or_first_config_ref_date(reference_dates)
+                                    for reference_dates in options['available_as_ofs'].values()])
+    options['current_date'] = options['initial_as_of']
 
     # set `models` and `initial_checked_models`
     options['models'] = []

src/hub_predtimechart/hub_config.py

@@ -4,6 +4,7 @@
 from pathlib import Path
 from typing import Optional
 
+import pandas as pd
 import yaml
 from jsonschema import validate
 from jsonschema.exceptions import ValidationError
@@ -122,6 +123,32 @@ def model_output_file_for_ref_date(self, model_id: str, reference_date: str) ->
         return None
 
 
+    def get_available_as_ofs(self) -> dict[str, list[str]]:
+        """
+        Returns a list of reference_dates with at least one forecast file.
+        """
+        # loop over every (reference_date X model_id) combination.
+        as_ofs = {self.fetch_target_id: set()}
+        for reference_date in self.reference_dates:  # ex: ['2022-10-22', '2022-10-29', ...]
+            for model_id in self.model_id_to_metadata:  # ex: ['Flusight-baseline', 'MOBS-GLEAM_FLUH', ...]
+                model_output_file = self.model_output_file_for_ref_date(model_id, reference_date)
+                if model_output_file:
+                    # todo xx extract to function, call from here and _generate_json_files()
+                    if model_output_file.suffix == '.csv':
+                        df = pd.read_csv(model_output_file, usecols=[self.target_col_name])
+                    elif model_output_file.suffix in ['.parquet', '.pqt']:
+                        df = pd.read_parquet(model_output_file, columns=[self.target_col_name])
+                    else:
+                        raise RuntimeError(f"unsupported model output file type: {model_output_file!r}. "
+                                           f"Only .csv and .parquet are supported")
+
+                    df = df.loc[df[self.target_col_name] == self.fetch_target_id, :]
+                    if not df.empty:
+                        as_ofs[self.fetch_target_id].add(reference_date)
+
+        return {fetch_target_id: sorted(list(reference_dates)) for fetch_target_id, reference_dates in as_ofs.items()}
+
+
 def _validate_hub_ptc_compatibility(ptc_config: dict, tasks: dict, model_metadata_schema: dict):
     """
     Validates a hub's predtimechart compatibility as identified in README.MD > Assumptions/limitations .

tests/expected/FluSight-forecast-hub/predtimechart-options.json

@@ -67,11 +67,10 @@
   "intervals": ["0%", "50%", "95%"],
   "initial_interval": "95%",
   "available_as_ofs": {
-    "wk inc flu hosp": [
-      "2023-10-07", "2023-10-14", "2023-10-21", "2023-10-28", "2023-11-04", "2023-11-11", "2023-11-18", "2023-11-25", "2023-12-02", "2023-12-09", "2023-12-16", "2023-12-23", "2023-12-30", "2024-01-06", "2024-01-13", "2024-01-20", "2024-01-27", "2024-02-03", "2024-02-10", "2024-02-17", "2024-02-24", "2024-03-02", "2024-03-09", "2024-03-16", "2024-03-23", "2024-03-30", "2024-04-06", "2024-04-13", "2024-04-20", "2024-04-27", "2024-05-04", "2024-05-11", "2024-11-16", "2024-11-23", "2024-11-30", "2024-12-07", "2024-12-14", "2024-12-21", "2024-12-28", "2025-01-04", "2025-01-11", "2025-01-18", "2025-01-25", "2025-02-01", "2025-02-08", "2025-02-15", "2025-02-22", "2025-03-01", "2025-03-08", "2025-03-15", "2025-03-22", "2025-03-29", "2025-04-05", "2025-04-12", "2025-04-19", "2025-04-26", "2025-05-03", "2025-05-10", "2025-05-17", "2025-05-24", "2025-05-31"]
+    "wk inc flu hosp": []
   },
-  "initial_as_of": "2025-05-31",
-  "current_date": "2025-05-31",
+  "initial_as_of": "2023-10-07",
+  "current_date": "2023-10-07",
   "models": [
     "CADPH-FluCAT_Ensemble", "CEPH-Rtrend_fluH", "CMU-TimeSeries", "CU-ensemble", "GT-FluFNP", "ISU_NiemiLab-NLH", "JHU_CSSE-CSSE_Ensemble", "LUcompUncertLab-chimera", "LosAlamos_NAU-CModel_Flu", "MIGHTE-Nsemble", "MOBS-GLEAM_FLUH", "NIH-Flu_ARIMA", "NU_UCSD-GLEAM_AI_FLUH", "PSI-PROF", "SGroup-RandomForest", "SigSci-CREG", "SigSci-TSENS", "Stevens-GBR", "UGA_flucast-Copycat", "UGA_flucast-INFLAenza", "UGuelph-CompositeCurve", "UGuelphensemble-GRYPHON", "UM-DeepOutbreak", "UMass-flusion", "UMass-trends_ensemble", "UNC_IDD-InfluPaint", "UVAFluX-Ensemble", "VTSanghani-Ensemble", "cfa-flumech", "cfarenewal-cfaepimlight", "fjordhest-ensemble"
   ],

tests/expected/covid19-forecast-hub/predtimechart-options.json

@@ -70,10 +70,10 @@
   "intervals": ["0%", "50%", "95%"],
   "initial_interval": "95%",
   "available_as_ofs": {
-    "wk inc covid hosp": ["2024-11-09", "2024-11-16", "2024-11-23", "2024-11-30", "2024-12-07", "2024-12-14", "2024-12-21", "2024-12-28", "2025-01-04", "2025-01-11", "2025-01-18", "2025-01-25", "2025-02-01", "2025-02-08", "2025-02-15", "2025-02-22", "2025-03-01", "2025-03-08", "2025-03-15", "2025-03-22", "2025-03-29", "2025-04-05", "2025-04-12", "2025-04-19", "2025-04-26", "2025-05-03", "2025-05-10", "2025-05-17", "2025-05-24", "2025-05-31"]
+    "wk inc covid hosp": []
   },
-  "initial_as_of": "2025-05-31",
-  "current_date": "2025-05-31",
+  "initial_as_of": "2024-11-09",
+  "current_date": "2024-11-09",
   "models": [],
   "initial_checked_models": [],
   "initial_xaxis_range": null

tests/expected/example-complex-forecast-hub/predtimechart-options.json

@@ -68,10 +68,10 @@
   "intervals": ["0%", "50%", "95%"],
   "initial_interval": "95%",
   "available_as_ofs": {
-    "wk inc flu hosp": ["2022-10-22", "2022-10-29", "2022-11-05", "2022-11-12", "2022-11-19", "2022-11-26", "2022-12-03", "2022-12-10", "2022-12-17", "2022-12-24", "2022-12-31", "2023-01-07", "2023-01-14", "2023-01-21", "2023-01-28", "2023-02-04", "2023-02-11", "2023-02-18", "2023-02-25", "2023-03-04", "2023-03-11", "2023-03-18", "2023-03-25", "2023-04-01", "2023-04-08", "2023-04-15", "2023-04-22", "2023-04-29", "2023-05-06", "2023-05-13", "2023-05-20", "2023-05-27"]
+    "wk inc flu hosp": ["2022-10-22", "2022-11-19", "2022-12-17"]
   },
-  "initial_as_of": "2023-05-27",
-  "current_date": "2023-05-27",
+  "initial_as_of": "2022-12-17",
+  "current_date": "2022-12-17",
   "models": ["Flusight-baseline", "MOBS-GLEAM_FLUH", "PSI-DICE"],
   "initial_checked_models": ["Flusight-baseline"],
   "disclaimer": "Most forecasts have failed to reliably predict rapid changes in the trends of reported cases and hospitalizations. Due to this limitation, they should not be relied upon for decisions about the possibility or timing of rapid changes in trends.",

tests/hub_predtimechart/test_hub_config.py

@@ -71,6 +71,14 @@ def test_model_output_file_for_ref_date():
     assert file is None
 
 
+def test_get_available_as_ofs():
+    hub_dir = Path('tests/hubs/example-complex-forecast-hub')
+    hub_config = HubConfig(hub_dir, hub_dir / 'hub-config/predtimechart-config.yml')
+    act_as_ofs = hub_config.get_available_as_ofs()
+    exp_as_ofs = {'wk inc flu hosp': ['2022-10-22', '2022-11-19', '2022-12-17']}
+    assert act_as_ofs == exp_as_ofs
+
+
 def test_hub_dir_existence():
     with pytest.raises(RuntimeError, match="hub_dir not found"):
         hub_dir = Path('tests/hubs/example-complex-forecast-hub')