Research

My research is broadly in spatial statistics and machine learning methodology, with a recurring interest in dependence structure — in point patterns, networks and time series — and in computational methods for inference when the likelihood is hard or impossible to evaluate directly. Recent work has combined this with applied machine learning, particularly for spatial and time-series data, following three years leading data science on energy-system modelling at a startup I co-founded.

Interests

• Machine learning methodology, including modern generative models and transformer architectures; application of machine learning to classification, prediction and data synthesis, particularly for spatial and time series data.
• Multivariate spatial point process models; characterisation of heterogeneous and anisotropic dependence structures; isotropy testing.
• Statistical modelling of network data, including community detection, clustering and online Bayesian changepoint detection.
• Likelihood-free inference, particularly approximate Bayesian computation in conjunction with Markov chain Monte Carlo and sequential Monte Carlo methods.

Selected research

• Multivariate Geometric Anisotropic Cox Processes

  Extending geometric anisotropy to multivariate spatial point processes, with an application to tropical forest ecology.