HGCN(O): A Self-Tuning GCN HyperModel Toolkit for Outcome Prediction in Event-Sequence Data
This work addresses outcome prediction in event-sequence data, such as for Predictive Business Process Monitoring, but appears incremental as it builds on existing GCN methods with toolkit optimizations.
The authors tackled the problem of predicting outcomes from event-sequence data by proposing HGCN(O), a self-tuning toolkit with four GCN architectures, which showed superior performance over traditional methods, with GCNConv models excelling on unbalanced datasets and all models performing consistently on balanced data.
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.