Gaussian processes for time-series modelling

Key Points

• Introduce the foundational conceptual framework of Bayesian non-parametric modelling using Gaussian processes for time-series data analysis.
• Reviewed the theoretical foundations of Bayesian non-parametric modelling applied to temporal data.
• Outlined strategies for integrating domain knowledge into Gaussian process model structure and covariance design.
• Applied Gaussian process workflows to illustrative case examples.
• Demonstrated that Gaussian processes provide a flexible, probabilistic approach for capturing complex patterns and uncertainties in temporal data.
• Showed that incorporating prior domain knowledge directly into model design enhances the interpretability and predictive performance of Gaussian processes across time-series applications.