ruptures

A Go re-implementation of change point detection algorithms, inspired by the Python library ruptures.

🚧 This project is in early development. APIs and module structure are subject to change.

✨ Overview

This library provides a modular implementations of change point detection algorithms such as:

• PELT (Pruned Exact Linear Time)
• Binary Segmentation
• Bottom-Up Segmentation
• Window-based approaches

It also includes cost functions (L1, L2, RBF, entropy) and metric evaluation utilities. The goal is to offer a fast, type-safe, and embeddable library in Go for time series segmentation and signal analysis.

📦 Features

• Idiomatic Go architecture
• Clean separation between cost functions and detection methods
• Fully compatible with custom input pipelines
• Suitable for CLI tools, web backends, or embedded systems

🔧 Project Layout

Estructura propuesta en Go, inspirada en la modularidad de ruptures:

rupture/
├── cmd/ # CLI para probar, correr, exportar resultados
├── core/
│ ├── base/ # Interfaces clave: CostFunction, Estimator
│ ├── cmdutils/ # │ ├── cost/ # Costos L1, L2, RBF... testables de forma independiente
│ ├── datasets/ # Simulación de señales (¡útil para tests!)
│ ├── detection/ # Algoritmos separados por carpeta = 💯
│ └── exceptions/ # `ErrInvalidSegment`, `ErrIncompatibleCost`, etc.
│ └── kernels/ # Base para libreria autonoma: pykernels port
│ └── linalg/ # utilerias de Gonum / Algebra lineal : shotcuts, envoltorios
│ └── stat/ # utilerias de Gonum / Stats: shotcuts, envoltorio + scypy portss
│ ├── metrics/ # Precision, coverage, F1, etc.
│ ├── types/ # │ ├── utils/ # Funciones como `SanityCheck()`, padding, slicing...
├── docs/ # Herramientas para generacion de documentacion / RAG
├── go.mod # Manejo de dependencias limpio
├── README.md # Main page
├── license # LICENSE
├── plan.md # Roadmap de implementación, dependencias, etc.
└── logbook.md # Registro de decisiones, experimentos o insights

🏁 Goals

• Port the core of ruptures to Go in idiomatic style
• Enable Go-native change point detection for scientific, industrial, and embedded applications
• Provide test coverage and validation against ruptures outputs
• Add statistical utilities and non-parametric tests in the long term

📚 Background

This project was inspired by ruptures, a Python library authored by Charles Truong and contributors, released under a BSD 3-Clause License.

While this project does not copy code from ruptures, it reimplements similar concepts and algorithmic structure. We gratefully acknowledge their work as the conceptual foundation of this library.

📄 License

See the LICENSE file for details.

🤝 Contributions

Contributions are welcome! If you're interested in joining the effort or adapting statistical algorithms to Go, feel free to open an issue or pull request.

---