EvaluationWindows.build#

classmethod EvaluationWindows.build(*, periods, dates, l_max, freq_offset, full_axis=False, include_carryover=True)[source]#

Work out the window of each period, and which of its dates are summed.

Parameters:
periodssequence of (pd.Timestamp, pd.Timestamp)

Period (start, end) pairs.

datespd.DatetimeIndex

All dates from the fitted data.

l_maxint

Evaluation-window half-length, already widened for any mediated path.

freq_offsetpd.DateOffset

Calendar-aware frequency offset.

include_carryoverbool, default=True

Whether the evaluation range reaches past the period to pick up its carry-out. Independent of the window, which always does: the two lengths do different jobs and collapsing them would change the numbers inside the period too.

full_axisbool, default=False

Evaluate every period on the complete fitted date axis, for an effect whose value depends on the whole series. Expressed as a window spanning every date rather than as a separate code path, so scenario building, input cutting and period aggregation stay identical: each period still perturbs only its own dates and still sums its increment over its own evaluation mask.

Returns:
EvaluationWindows

The windows, and the length they stack to.