Manifold Learning with Approximate Nearest Neighbors including on statistical manifolds.
The theory of probabilistic reconciliation and score optimisation by gradient descent.
Bayesian methods taking survey weights into account are proposed, evaluated and applied to estimating the Australian income distribution.
Application of hierarchical methods to Australian GDP Forecasting.
The geometric understanding of hierarchical time series and forecast reconciliation.
Forecast Reconciliation (with a novel Bayesian approach) applied to Swiss Export Data.
New algorithms for probabilistic forecast reconciliation. Application is to energy data following temporal hierarchies.
Introduces two Variational Bayes algorithms tailored for online inference. Applied to predicting the location of vehicles.
Forecast reconciliation used for mortality projections broken down by cause-of-death.
Forecasting Australian macroeconomic variables using popular regularisation techniques.