HGCN(O): A Self-Tuning GCN HyperModel Toolkit for Outcome Prediction in Event-Sequence Data

TL;DR

HGCN(O) is a self-tuning toolkit that leverages multiple GCN architectures and graph representations to improve event sequence prediction accuracy and stability, especially in unbalanced datasets, with applications in business process monitoring.

Contribution

The paper introduces HGCN(O), a novel self-tuning GCN-based toolkit with four architectures for enhanced event sequence prediction in diverse datasets.

Findings

GCNConv models perform better on unbalanced data

All models show consistent performance on balanced data

HGCN(O) outperforms traditional prediction approaches

Abstract

We propose HGCN(O), a self-tuning toolkit using Graph Convolutional Network (GCN) models for event sequence prediction. Featuring four GCN architectures (O-GCN, T-GCN, TP-GCN, TE-GCN) across the GCNConv and GraphConv layers, our toolkit integrates multiple graph representations of event sequences with different choices of node- and graph-level attributes and in temporal dependencies via edge weights, optimising prediction accuracy and stability for balanced and unbalanced datasets. Extensive experiments show that GCNConv models excel on unbalanced data, while all models perform consistently on balanced data. Experiments also confirm the superior performance of HGCN(O) over traditional approaches. Applications include Predictive Business Process Monitoring (PBPM), which predicts future events or states of a business process based on event logs.