deepcharles/ruptures

What it solves

It provides tools for off-line change point detection, allowing users to analyze and segment non-stationary signals to identify where the underlying properties of the data change over time.

How it works

The library implements various parametric and non-parametric models to detect change points. It offers both exact and approximate detection algorithms and uses a modular structure that allows different models and algorithms to be connected and extended.

Who it’s for

It is designed for researchers and practitioners working with signal processing and time-series analysis across diverse fields, including remote sensing (e.g., crop monitoring), biology (e.g., neurotransmitter regulation), sports science, and crystallography.

Highlights

• Consistent and well-documented interface for ease of use.
• Supports both exact and approximate detection methods.
• Modular architecture for extending algorithms and models.
• Proven utility in real-world applications like NASA crop monitoring and marine biodiversity observation.