tseda.change_point.change_point_estimator module

Change-point detection utilities based on the PELT algorithm.

class tseda.change_point.change_point_estimator.PELT_ChangePointEstimator(series: pandas.Series, model: str = 'rbf')

Bases: object

Estimate change points with the PELT algorithm and return a predicted segment series.

Initialize PELT estimator and fit the algorithm on the input series.

Parameters:

• series – Non-empty numeric series indexed by datetimes.

• model – Cost model used by PELT (default "rbf").

__init__(series: pandas.Series, model: str = 'rbf') → None

Initialize PELT estimator and fit the algorithm on the input series.

Parameters:

• series – Non-empty numeric series indexed by datetimes.

• model – Cost model used by PELT (default "rbf").

predict_series() → pandas.Series

Return the predicted segment label series.

Returns:

Series with the same index as the input series and values such as "segment-1", "segment-2", etc.

class tseda.change_point.change_point_estimator.ChangePointEstimator(series: pandas.Series)

Bases: object

Compatibility wrapper used by existing tests and call sites.

Initialize the estimator with the input series.

Parameters:

series – Non-empty numeric series indexed by datetimes.

__init__(series: pandas.Series) → None

Initialize the estimator with the input series.

Parameters:

series – Non-empty numeric series indexed by datetimes.

estimate_change_points(penalty_coeff: float = 2.0) → pandas.Series

Run PELT and assign segment labels for each observation.

Parameters:

penalty_coeff – Multiplier applied to log(n) when building the PELT penalty term.

Returns:

Series of segment labels aligned to the original input index.