We are delighted to announce the upcoming Compass Guest Lecture with Dr Vincenzo Gioia (University of Trieste) and Professor Ruggero Bellio (University of Udine).

Schedule

11am – 12pm: Scalable Estimation of Probit Models with Crossed Random Effects, Professor Ruggero Bellio

1 – 2pm: Additive Covariance Matrix Models: Modelling Regional Electricity Net-Demand in Great Britain, Dr Vincenzo Gioia

Abstracts

Scalable Estimation of Probit Models with Crossed Random Effects
Professor Ruggero Bellio, Department of Economics and Statistics, University of Udine (Italy)

This talk illustrates a scalable approach to mixed effects modeling with a probit link and a crossed random effects error structure. Random effects with a crossed structure arise often in social and business applications, a notable setting being that of electronic commerce, with random effects related to customers and purchased items, respectively. In sparsely sampled crossed data the computation for both frequentist and Bayesian estimation can easily grow superlinearly with respect to the sample size, which severely limits the use of these models for very large settings. The proposed method belongs to the class of composite likelihood estimators, and entails the fit of three misspecified reduced models. The resulting estimator is consistent and has an overall computational cost linear in the number of observations. This is a joint work with Art Owen and Swarnadip Ghosh, Stanford University, and Cristiano Varin, Ca’Foscari University of Venice.

Additive Covariance Matrix Models: Modelling Regional Electricity Net-Demand in Great Britain
Dr Vincenzo Gioia, Department of Economics, Business, Mathematics and Statistics University of Trieste (Italy)

Forecasts of regional electricity net-demand, consumption minus embedded generation, are an essential input for reliable and economic power system operation, and energy trading. While such forecasts are typically performed region by region, operations such as managing power flows require spatially coherent joint forecasts, which account for cross-regional dependencies. Here we forecast the joint distribution of net demand across the 14 regions constituting Great Britain’s electricity network. Joint modelling is complicated by the fact that the net-demand variability within each region, and the dependencies between regions, vary with temporal, socio-economical and weather-related factors. We accommodate for these characteristics by proposing a multivariate Gaussian model based on a modified Cholesky parametrisation, which allows us to model each unconstrained parameter via an additive model. Given that the number of model parameters and covariates is large, we adopt a semi-automated approach to model selection, based on gradient boosting. In addition to demonstrating that adopting a covariate-dependent covariance matrix model leads to substantial forecasting performance improvements, comparable to those obtained by using a full rather than a diagonal static covariance matrix, we explore the model output via accumulated local effects and other visual tools to get insights into how the covariates affect net-demand variability and dependencies. This is a joint work with Matteo Fasiolo, University of Bristol, Jethro Browell, University of Glasgow, and Ruggero Bellio, University of Udine.