Window

class openstef_beam.evaluation.Window(lag: timedelta, size: timedelta, stride: timedelta = timedelta(days=1), minimum_coverage: float = 0.5)

Bases: PydanticStringPrimitive

Represents a time window with lag, size, and stride parameters.

Used for defining sliding windows for time series analysis with: - lag: How far back from the reference point to start the window - size: The duration of the window - stride: How much to advance for the next window - minimum_coverage: Minimum required data coverage (0.0-1.0)

Parameters:

• lag (timedelta)

• size (timedelta)

• stride (timedelta)

• minimum_coverage (float)

__init__(lag: timedelta, size: timedelta, stride: timedelta = timedelta(days=1), minimum_coverage: float = 0.5)

Initialize a time window with the specified parameters.

Parameters:

• lag (timedelta) – How far back from the reference point to start the window.

• size (timedelta) – The duration of the window.

• stride (timedelta) – How much to advance for the next window.

• minimum_coverage (float) – Minimum required data coverage (0.0-1.0).

• lag

• size

• stride

• minimum_coverage

__str__() → str

Converts to string in ‘(lag=X,size=Y,stride=Z)’ format.

Timedeltas are serialized in ISO 8601 format.

Return type:

str

Returns:

String representation of the window parameters.

classmethod from_string(s: str) → Self

Creates an instance from a string in ‘(lag=X,size=Y,stride=Z)’ format.

Parameters:

• s (str) – String representation to parse.

• s

Returns:

Window instance parsed from the string.

Raises:

ValueError – If the string format is invalid.

Return type:

Self

classmethod validate(v: Self | str | dict[str, Any], _info: Any = None) → Self

Validates and converts various input types to Window.

Parameters:

• v (Union[Self, str, dict[str, Any]]) – Input value to validate (Window, string, or dict).

• _info (Any) – Additional validation info (unused).

• v

• _info

Returns:

Validated Window instance.

Return type:

Self