Core API overview

The tsseg package provides a thin façade over several namespaces that are described throughout the guides. This page recaps the most relevant entry points and links to the narrative documentation where the full usage examples live.

Segments and detectors

tsseg.algorithms exposes detector classes for two main tasks:

Change point detection — AMOC, BinSeg, BOCD, BottomUp, ClaSP, DynP, E-Agglo, ESPRESSO, FLUSS, GGS, iCID, IGTS, KCPD, PELT, Prophet, TIRE, TS-CP2, tGLAD, Window.

State detection — AutoPlait, CLaP, E2USD, HDP-HSMM, Hidalgo, HMM, PaTSS, TICC, Time2State, VQ-TSS, VSAX.

A Random baseline is also included as a lower bound for benchmarks.

Each detector follows the standard fit/predict/fit_predict contract documented in Detectors overview.

Datasets

tsseg.data.datasets bundles a small collection of datasets (e.g. MoCap) plus convenience functions for loading data. Refer to Datasets for details on file formats and helper functions.

Evaluation

tsseg.metrics contains scoring helpers for both change-point detection and state-labelling evaluation. The most common utilities are summarised in Metric helpers.