preprocessing.coverage.assert_actual_lag_within

preprocessing.coverage.assert_actual_lag_within(actual, now, max_lag)

Raise CoverageError if the last published actual is too old.

Corresponds to the second guard in the operational assert_coverage (script line ~462): the last non-NaN Actual Load observation is older than now - max_lag, indicating ENTSO-E’s publication pipeline has stalled.

The comparison is strict (<): last_actual == now - max_lag is a passing boundary (lag is exactly at the tolerance).

Parameters

Name	Type	Description	Default
actual	pd.Series	Series of Actual Load (or equivalent) values. NaNs are stripped before computing the last published timestamp.	required
now	pd.Timestamp	Reference timestamp representing the current wall-clock time (tz-aware).	required
max_lag	pd.Timedelta	Maximum acceptable age of the last published observation. The operational default is `pd.Timedelta(hours=36)`.	required

Raises

Name	Type	Description
	CoverageError	When the last non-NaN observation is older than `now - max_lag`, reporting the observed lag in whole hours.

Examples

import pandas as pd
from spotforecast2_safe.preprocessing.coverage import assert_actual_lag_within
from spotforecast2_safe.exceptions import CoverageError

now = pd.Timestamp("2026-06-11 12:00", tz="UTC")
idx = pd.date_range("2026-06-10 00:00", periods=24, freq="h", tz="UTC")
actual = pd.Series(range(24), index=idx, dtype=float)

# Passing: last observation is 36 h old, tolerance is exactly 36 h.
assert_actual_lag_within(actual, now, pd.Timedelta(hours=36))

# Failing: tolerance is 35 h but last observation is 36 h old.
try:
    assert_actual_lag_within(actual, now, pd.Timedelta(hours=35))
except CoverageError as exc:
    print(exc)