HGCN(O): A Self-Tuning GCN HyperModel Toolkit for Outcome Prediction in Event-Sequence Data
Researchers have introduced HGCN(O), a self-tuning toolkit designed for predicting outcomes in event-sequence data using Graph Convolutional Networks (GCNs). The toolkit incorporates four distinct GCN architectures and allows for various graph representations, optimizing for both prediction accuracy and stability across balanced and unbalanced datasets. Experiments demonstrate that HGCN(O) outperforms traditional methods, with specific GCNConv models showing particular strength on unbalanced data, and finds applications in areas like Predictive Business Process Monitoring. AI
IMPACT Enhances predictive capabilities for event-driven systems, potentially improving business process monitoring and similar applications.