Event Log Sampling for Predictive Monitoring

TL;DR

This paper introduces an instance sampling method for predictive process monitoring that accelerates training of machine learning models while preserving prediction accuracy, benefiting process stakeholders.

Contribution

It presents a novel sampling technique that improves training efficiency in predictive process monitoring without sacrificing accuracy.

Findings

Sampling speeds up training significantly

Prediction accuracy remains reliable

Method is effective for next activity prediction

Abstract

Predictive process monitoring is a subfield of process mining that aims to estimate case or event features for running process instances. Such predictions are of significant interest to the process stakeholders. However, state-of-the-art methods for predictive monitoring require the training of complex machine learning models, which is often inefficient. This paper proposes an instance selection procedure that allows sampling training process instances for prediction models. We show that our sampling method allows for a significant increase of training speed for next activity prediction methods while maintaining reliable levels of prediction accuracy.