In analytics, a significant source of information stems from data collected over time. These time-series data might be event sequences, aggregated event sequences (e.g., sales data), or measurements of physical quantities (sensor data). In typical analytics applications, at least two of these different types of time-series data are combined with univariate attributes to gain insights into complex systems and guide decision-making.
Consequently, three major challenges have to be solved frequently for data-driven models in time-series analytics:
- The models need to be robust against disproportional availability of different time-series data types.
- The models need to be interpretable to support the decision-making of domain experts.
- The output of the data-driven models needs to be combined with decision models.
This collection focuses on novel approaches and recent applications of time-series analytics to address these challenges. Contributions on applied time-series machine learning, interpretable time-series modelling, causal time-series inference, non-stationary sequence models, Bayesian temporal models, automated time-series feature engineering, systematic time-series feature engineering, time-series engineering, normality models, simulation frameworks, and big data, small sample problems are welcomed but are not restricted to these topics.
Applications might range from predictive maintenance, health care, business and economics, engineering sciences, to novel application fields.
Submission Deadline is 26 November 2026.