Event Time Prediction
Overview
Event time prediction studies when an event is likely to happen, from the next interaction in a network to a clinical milestone or equipment failure, helping people plan and intervene. Common approaches include survival models, temporal point processes such as Hawkes models, and sequence or Transformer models that combine event histories with contextual data; some also predict event type and timing jointly. Researchers evaluate these methods on settings such as electronic health records, business processes, transportation, and sensor data, using measures of timing error, ranking, and probabilistic calibration. A central challenge is that event times can be irregular, censored, heavy-tailed, or inherently uncertain, and complex models do not always outperform simpler baselines or generalize across settings.
Latest papers
47 papers6 papers in the last four weeks, up 20% on the four weeks before. 0.0% of all new papers.