Scalable Marked Point Processes for Exchangeable and Non-Exchangeable Event Sequences

TL;DR

This paper introduces a scalable, interpretable framework for marked point processes that effectively models both exchangeable and non-exchangeable event sequences using variational inference, without extensive tuning or pre-training.

Contribution

The authors develop a general, scalable inferential framework inspired by Hawkes processes that handles diverse event sequences with minimal tuning and no pre-training, outperforming existing methods.

Findings

Competitive computational performance demonstrated on real data

Effective modeling of both exchangeable and non-exchangeable sequences

Successful large-scale application in sports event analysis

Abstract

We adopt the interpretability offered by a parametric, Hawkes-process-inspired conditional probability mass function for the marks and apply variational inference techniques to derive a general and scalable inferential framework for marked point processes. The framework can handle both exchangeable and non-exchangeable event sequences with minimal tuning and without any pre-training. This contrasts with many parametric and non-parametric state-of-the-art methods that typically require pre-training and/or careful tuning, and can only handle exchangeable event sequences. The framework's competitive computational and predictive performance against other state-of-the-art methods are illustrated through real data experiments. Its attractiveness for large-scale applications is demonstrated through a case study involving all events occurring in an English Premier League season.