Working My Way Back to You: Resource-Centric Next-Activity Prediction

2025-08-26

AI Summary

Most existing next-activity prediction studies focus solely on control-flow patterns, neglecting the resource perspective. This paper pioneers a systematic investigation of next-activity prediction from a resource-centric viewpoint, proposing a novel paradigm grounded in individual resource behavior modeling—rather than case-level process modeling—to support workforce optimization, load balancing, and capacity forecasting. Methodologically, we design a resource encoding strategy integrating 2-gram transition patterns and activity repetition features, and comparatively evaluate three encoding variants across LightGBM, Transformer, and Random Forest models. Experiments on four real-world event logs demonstrate that the 2-gram encoding significantly improves performance for LightGBM and Transformer; the fused encoding substantially boosts Random Forest accuracy; and the overall best-performing configuration achieves the highest mean accuracy, empirically confirming that explicit resource information delivers substantial predictive gains.

Technology Category

• Planning, Routing, and Scheduling: Activity and Plan Recognition
• Search and Optimization: Metareasoning and Metaheuristics
• Knowledge Representation and Reasoning: Action, Change, and Causality

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Web Mining and Content Analysis: Models for Web evolution

Abstract

Predictive Process Monitoring (PPM) aims to train models that forecast upcoming events in process executions. These predictions support early bottleneck detection, improved scheduling, proactive interventions, and timely communication with stakeholders. While existing research adopts a control-flow perspective, we investigate next-activity prediction from a resource-centric viewpoint, which offers additional benefits such as improved work organization, workload balancing, and capacity forecasting. Although resource information has been shown to enhance tasks such as process performance analysis, its role in next-activity prediction remains unexplored. In this study, we evaluate four prediction models and three encoding strategies across four real-life datasets. Compared to the baseline, our results show that LightGBM and Transformer models perform best with an encoding based on 2-gram activity transitions, while Random Forest benefits most from an encoding that combines 2-gram transitions and activity repetition features. This combined encoding also achieves the highest average accuracy. This resource-centric approach could enable smarter resource allocation, strategic workforce planning, and personalized employee support by analyzing individual behavior rather than case-level progression. The findings underscore the potential of resource-centric next-activity prediction, opening up new venues for research on PPM.

Problem

• Predicting next activities from resource perspective
• Evaluating models and encodings for resource-centric prediction
• Improving resource allocation and workforce planning strategies

Innovation

• Resource-centric viewpoint for next-activity prediction
• LightGBM and Transformer models with 2-gram encoding
• Combined encoding strategy for Random Forest optimization