Remove old rolling viewpoint forecasting infrastructure

documentation/changelog.rst

@@ -17,6 +17,7 @@ New features
 
 Infrastructure / Support
 ----------------------
+* Remove old rolling viewpoint forecasting infrastructure: ``make_fixed_viewpoint_forecasts`` (which only raised ``NotImplementedError``), ``make_rolling_viewpoint_forecasts`` (renamed to ``make_forecasts``), and the ``supported_horizons`` utility. The invalid-horizon validation check is also removed [see `PR #2075 <https://www.github.com/FlexMeasures/flexmeasures/pull/2075>`_]
 * Stop creating new toy assets when restarting the docker-compose stack [see `PR #2018 <https://www.github.com/FlexMeasures/flexmeasures/pull/2018>`_]
 * Migrate from ``pip`` to ``uv`` for dependency management, and from ``make`` to ``poe`` [see `PR #1973 <https://github.com/FlexMeasures/flexmeasures/pull/1973>`_]
 * Improve contact information to get in touch with the FlexMeasures community [see `PR #2022 <https://www.github.com/FlexMeasures/flexmeasures/pull/2022>`_]

flexmeasures/data/models/forecasting/exceptions.py

@@ -1,6 +1,2 @@
 class NotEnoughDataException(Exception):
     pass
-
-
-class InvalidHorizonException(Exception):
-    pass

flexmeasures/data/services/forecasting.py

@@ -16,7 +16,6 @@
 
 from flexmeasures.data import db
 from flexmeasures.data.models.forecasting import lookup_model_specs_configurator
-from flexmeasures.data.models.forecasting.exceptions import InvalidHorizonException
 from flexmeasures.data.models.time_series import Sensor, TimedBelief
 from flexmeasures.data.models.forecasting.utils import (
     get_query_window,
@@ -27,13 +26,12 @@
     as_server_time,
     server_now,
     forecast_horizons_for,
-    supported_horizons,
 )
 
 """
 The life cycle of a forecasting job:
 1. A forecasting job is born in create_forecasting_jobs.
-2. It is run in make_rolling_viewpoint_forecasts or make_fixed_viewpoint_forecasts, which write results to the db.
+2. It is run in make_forecasts, which writes results to the db.
    This is also where model specs are configured and a possible fallback model is stored for step 3.
 3. If an error occurs (and the worker is configured accordingly), handle_forecasting_exception comes in.
    This might re-enqueue the job or try a different model (which creates a new job).
@@ -96,7 +94,7 @@ def create_forecasting_jobs(
     jobs: list[Job] = []
     for horizon in horizons:
         job = Job.create(
-            make_rolling_viewpoint_forecasts,
+            make_forecasts,
             kwargs=dict(
                 sensor_id=sensor_id,
                 horizon=horizon,
@@ -130,35 +128,18 @@ def create_forecasting_jobs(
     return jobs
 
 
-def make_fixed_viewpoint_forecasts(
+def make_forecasts(
     sensor_id: int,
     horizon: timedelta,
     start: datetime,
     end: datetime,
     custom_model_params: dict = None,
 ) -> int:
-    """Build forecasting model specs, make fixed-viewpoint forecasts, and save the forecasts made.
+    """Build forecasting model specs, make forecasts, and save the forecasts made.
 
     Each individual forecast is a belief about a time interval.
-    Fixed-viewpoint forecasts share the same belief time.
-    See the timely-beliefs lib for relevant terminology.
-    """
-    # todo: implement fixed-viewpoint forecasts
-    raise NotImplementedError
-
-
-def make_rolling_viewpoint_forecasts(
-    sensor_id: int,
-    horizon: timedelta,
-    start: datetime,
-    end: datetime,
-    custom_model_params: dict = None,
-) -> int:
-    """Build forecasting model specs, make rolling-viewpoint forecasts, and save the forecasts made.
-
-    Each individual forecast is a belief about a time interval.
-    Rolling-viewpoint forecasts share the same belief horizon (the duration between belief time and knowledge time).
-    Model specs are also retrained in a rolling fashion, but with its own frequency set in custom_model_params.
+    Forecasts share the same belief horizon (the duration between belief time and knowledge time).
+    Model specs are retrained periodically, with frequency set in custom_model_params.
     See the timely-beliefs lib for relevant terminology.
 
     Parameters
@@ -214,12 +195,7 @@ def make_rolling_viewpoint_forecasts(
     rq_job.meta["fallback_model_search_term"] = fallback_model_search_term
     rq_job.save()
 
-    # before we run the model, check if horizon is okay and enough data is available
-    if horizon not in supported_horizons():
-        raise InvalidHorizonException(
-            "Invalid horizon on job %s: %s" % (rq_job.id, horizon)
-        )
-
+    # before we run the model, check if enough data is available
     query_window = get_query_window(
         model_specs.start_of_training,
         end,
@@ -290,7 +266,7 @@ def handle_forecasting_exception(job, exc_type, exc_value, traceback):
     if "fallback_model_search_term" in job.meta:
         if job.meta["fallback_model_search_term"] is not None:
             new_job = Job.create(
-                make_rolling_viewpoint_forecasts,
+                make_forecasts,
                 args=job.args,
                 kwargs=job.kwargs,
                 connection=current_app.queues["forecasting"].connection,

flexmeasures/data/tests/test_forecasting_jobs.py

@@ -199,25 +199,6 @@ def test_failed_forecasting_insufficient_data(
     check_failures(app.queues["forecasting"], 2 * ["NotEnoughDataException"])
 
 
-def test_failed_forecasting_invalid_horizon(
-    app, run_as_cli, clean_redis, setup_test_data
-):
-    """This one (as well as the fallback) should fail as the horizon is invalid."""
-
-    # asset has only 1 power sensor
-    solar_device_1: Sensor = setup_test_data["solar-asset-1"].sensors[0]
-
-    create_forecasting_jobs(
-        start_of_roll=as_server_time(datetime(2015, 1, 1, 21)),
-        end_of_roll=as_server_time(datetime(2015, 1, 1, 23)),
-        horizons=[timedelta(hours=18)],
-        sensor_id=solar_device_1.id,
-        custom_model_params=custom_model_params(),
-    )
-    work_on_rq(app.queues["forecasting"], exc_handler=handle_forecasting_exception)
-    check_failures(app.queues["forecasting"], 2 * ["InvalidHorizonException"])
-
-
 def test_failed_unknown_model(app, clean_redis, setup_test_data):
     """This one should fail because we use a model search term which yields no model configurator."""
 

flexmeasures/utils/time_utils.py

@@ -279,15 +279,6 @@ def forecast_horizons_for(resolution: str | timedelta) -> list[str] | list[timed
         return horizons
 
 
-def supported_horizons() -> list[timedelta]:
-    return [
-        timedelta(hours=1),
-        timedelta(hours=6),
-        timedelta(hours=24),
-        timedelta(hours=48),
-    ]
-
-
 def timedelta_to_pandas_freq_str(resolution: timedelta) -> str:
     return to_offset(resolution).freqstr