Tick is a Python 3 library for statistical learning, specializing in time-dependent modeling for applications like point processes, generalized linear models, and optimization. It targets researchers and engineers working with temporal data, offering tools for inference, simulation, and efficient computation on various hardware.

How It Works

Tick's core is an optimization module featuring computational classes, solvers, and proximal operators for regularization. This design enables efficient handling of complex statistical models, particularly those involving time-series data and point processes. It provides tools for simulating and inferring Hawkes models with various kernel assumptions, facilitating advanced analysis of event sequences.

Quick Start & Requirements

• Install: pip install tick
• Requirements: Python 3.5+, Linux/macOS (Windows experimental). Dependencies are managed by pip.
• Documentation: https://x-datainitiative.github.io/tick
• Examples: https://x-datainitiative.github.io/tick/auto_examples/index.html

Highlighted Details

• Specializes in time-dependent modeling, including point processes and Hawkes models.
• Features a generic optimization toolbox with proximal operators for regularization.
• Optimized using Intel® Math Kernel Library (MKL) for performance on Intel processors.
• Used in industrial applications for pharmacovigilance, financial order book modeling, and social media analysis.

Maintenance & Community

• Initiated at École Polytechnique, France.
• Community channel available on Gitter: https://gitter.im/xdata-tick
• Bug reporting via GitHub issues.

Licensing & Compatibility

• Distributed under the 3-Clause BSD license.
• Permissive license suitable for commercial use and integration into closed-source projects.

Limitations & Caveats

Windows support is experimental. The library was last updated in 2017, indicating potential for unaddressed issues or lack of recent feature development.