Zammit-Mangion, Andrew Dr

Senior Lecturer

  • National Institute for Applied Statistical Research Australia
  • Faculty of Engineering and Information Sciences
  • School of Mathematics and Applied Statistics

Overview

Professional Activities:

Member of the American Statistical Association and the Institute of Electrical and Electronics Engineers

Awards:

04.2013: National Academy of Sciences (NAS) of the USA Cozzarelli Prize for paper of outstanding excellence and originality, WDC, USA.
11.2012: Institute of Engineering and Technology (IET) Control and Automation Best Doctoral Dissertation Prize, London, UK.
10.2008: Best Student Paper at the UK Automatic Control Council (UKACC) Conference, Manchester, UK.
04.2008: Full University Scholarship for pursuing a PhD at the University of Sheffield, UK.
07.2007: RS Prize for Best Academic Achievement at the Faculty of Engineering, University of Malta.

Top Publications

Research Overview

  • My research interests lie in spatial and spatio-temporal modelling and the tools that enable it. In previous work, I have focused on variational Bayesian methods for approximate inference of spatio-temporal log-Gaussian Cox process models. These methods were successfully applied in conflict modelling and, in more recent research, I am working on approximate message passing algorithms in this context. In other related previous work, I have used well-established approximations to spatio-temporal multivariate processes to assess the Antarctic contribution to sea-level rise. For project details please see here. The project involved fusing multiple data products (from diverse satellites and research groups) through the use of a large-scale spatio-temporal model. Work here involved the use of the message-passing interface on a high-performance computer, parallel Gibbs sampling methods, sparse-matrix methodologies and has resulted in over five publications to date (for the latest publication see here).

    My recent work at NIASRA has focused on multivariate spatial modelling and atmospheric trace-gas inversion. I have taken particular interest in this area since it forces me to move beyong the usual 'Gaussian' spatial models, it is inter-disciplinary, and requires an eye for careful computation. In a similar vein to my previous work on Antarctica, the work has extremely important implications, allowing one to quantify where the biggest flux sources and sinks are simply from observations of gas concentrations. Lately, I have also taken on reproducibility and software contribution to the open-source community more actively and have written a number of reproducible packages intended solely to reproduce the results in recent papers (see here and here) as well some intended for use by the general scientific community (see here and here). The latter one, focused on Fixed-Rank Kriging, is still under intense development.

Available as Research Supervisor

Selected Publications

Available as Research Supervisor

Advisees

  • Graduate Advising Relationship

    Degree Research Title Advisee
    Doctor of Philosophy Data Privacy and Data Utility - Data Masking Schemes and Their Applications Wakefield, Bradley
    Doctor of Philosophy Emulation For Spatio-Temporal Sensitivities With Application To Atmospheric Transport Models Cartwright, Laura
    Doctor of Philosophy Deep statistical models with applications to environmental data Vu, Dang Minh Quan

Teaching Overview

  • Lecturer in Generalised Linear Models (STAT332, 2015), Applied Probability and Financial Risk (STAT304, 2017). and Estimation and Hypothesis Testing (STAT232, 2017).

Full Name

  • Andrew Zammit-Mangion

Mailing Address

  • 39C Northfields Avenue

    University of Wollongong

    Wollongong

    NSW

    2522

    Australia

Top Publications

Research Overview

  • My research interests lie in spatial and spatio-temporal modelling and the tools that enable it. In previous work, I have focused on variational Bayesian methods for approximate inference of spatio-temporal log-Gaussian Cox process models. These methods were successfully applied in conflict modelling and, in more recent research, I am working on approximate message passing algorithms in this context. In other related previous work, I have used well-established approximations to spatio-temporal multivariate processes to assess the Antarctic contribution to sea-level rise. For project details please see here. The project involved fusing multiple data products (from diverse satellites and research groups) through the use of a large-scale spatio-temporal model. Work here involved the use of the message-passing interface on a high-performance computer, parallel Gibbs sampling methods, sparse-matrix methodologies and has resulted in over five publications to date (for the latest publication see here).

    My recent work at NIASRA has focused on multivariate spatial modelling and atmospheric trace-gas inversion. I have taken particular interest in this area since it forces me to move beyong the usual 'Gaussian' spatial models, it is inter-disciplinary, and requires an eye for careful computation. In a similar vein to my previous work on Antarctica, the work has extremely important implications, allowing one to quantify where the biggest flux sources and sinks are simply from observations of gas concentrations. Lately, I have also taken on reproducibility and software contribution to the open-source community more actively and have written a number of reproducible packages intended solely to reproduce the results in recent papers (see here and here) as well some intended for use by the general scientific community (see here and here). The latter one, focused on Fixed-Rank Kriging, is still under intense development.

Selected Publications

Advisees

  • Graduate Advising Relationship

    Degree Research Title Advisee
    Doctor of Philosophy Data Privacy and Data Utility - Data Masking Schemes and Their Applications Wakefield, Bradley
    Doctor of Philosophy Emulation For Spatio-Temporal Sensitivities With Application To Atmospheric Transport Models Cartwright, Laura
    Doctor of Philosophy Deep statistical models with applications to environmental data Vu, Dang Minh Quan

Teaching Overview

  • Lecturer in Generalised Linear Models (STAT332, 2015), Applied Probability and Financial Risk (STAT304, 2017). and Estimation and Hypothesis Testing (STAT232, 2017).

Full Name

  • Andrew Zammit-Mangion

Mailing Address

  • 39C Northfields Avenue

    University of Wollongong

    Wollongong

    NSW

    2522

    Australia

Visualisations

  Publication History
 

Research Areas

Geographic Focus