hawkeslib 0.2.2

parameter estimation for simple Hawkes (self-exciting) processes

Welcome to hawkeslib

hawkeslib started with the ambition of having a simple Python implementation of plain-vanilla Hawkes (or self-exciting processes), i.e. those with factorized triggering kernels with exponential decay functions.

The docs contain tutorials, examples and a detailed API reference. For other examples, see the examples/ folder.

The following models are available:

• Univariate Hawkes Process (with exponential delay)
• Bayesian Univariate Hawkes Process (with exponential delay)
• Poisson Process
• 'Bayesian' Poisson process

Bayesian variants implement methods for sampling from the posterior as well as calculating marginal likelihood (e.g. for Bayesian model comparison).

Installation

Cython (>=0.28) and numpy (>=1.14) and scipy must be installed prior to the installation as they are required for the build.

$ pip install -U Cython numpy scipy
$ pip install hawkeslib