tsdive.segment ¶

segment(
    archive: str | Path,
    window: str | None = None,
    *,
    penalty: float | None = None,
    min_size: int = 10,
    basis: str
    | CalculationBasis = CalculationBasis.TIME_WEIGHTED,
    stepped: bool = False,
) -> SegmentAnalysis

Segment one window of a single-tag archive into regimes.

Parameters:

Name	Type	Description	Default
`archive`	`str \| Path`	parquet archive carrying `tsdive.meta`.	required
`window`	`str \| None`	`START/END` in ISO 8601 UTC. Omitted, the archive's own extent.	`None`
`penalty`	`float \| None`	penalty added for each breakpoint. Omitted, the default multiplier times `log(n)` on the scaled series.	`None`
`min_size`	`int`	samples a segment must hold, at least.	`10`
`basis`	`str \| CalculationBasis`	calculation basis declared on the read.	`TIME_WEIGHTED`
`stepped`	`bool`	stepped interpolation between samples.	`False`

Raises:

Type	Description
`ValueError`	malformed or naive window.
`TSDiveError`	any typed refusal from the read path.

Examples:

>>> import tsdive
>>> s = tsdive.segment("data/demo/fic101_demo.parquet")
>>> len(s.found.breakpoints), len(s.frame)
(5, 6)
>>> list(s.frame.columns)
['index', 'start', 'end', 'n', 'median', 'mad']