Skip to Content
Scan the QR code using the NZ COVID Tracer app when you enter campus.
I am a statistician with a background predominately in the field of transportation science, as well as the field of applied multivariate statistics.
Originally, I studied Statistik at Dortmund Universität (Germany) before doing my Masters in Statistics at the University of Auckland (2009), and then my PhD in Statistics at Massey Univeristy (2013). After completing my doctoral work, I joined Auckland University of Technology from 2012 to 2016 as a lecturer in the Mathematical and Computation Sciences. I then returned to Massey University as a lecturer in Statistics in 2016.
My research in transportation studies naturally fall into three interwoven themes.
The first involves developing methods of statistical inference that address the issue of indeterminacy. This is a fundamental problem in transportation because the data available, traffic link counts, does not uniquely describe the ways in which people travel through a system. My work therefore focuses on a complex class of statistical inverse problems. A technique using a Bayesian MCMC framework was only available for very simple linear networks, such as one-way bus routes. I expanded the technique to apply to larger networks with bifurcations.
My second research theme relates to how existing transportation models struggle to recreate the kind of day-to-day dynamics we observe in real-life traffic flows. I developed a model that includes additional temporal variation in a parsimonious manner by redefining some of the fixed variables as random variables. This new class of doubly-stochastic process model, is able to produce simulated flows that display the kind of non-stationary behaviour that we observe in real data.
Finally, these models are very complex and the general question arises on how to well simple data such as link counts function as a means of making inference. Being able to compare models to each other within the Bayesian framework using DIC has allowed me to deliver evidence, of the usefulness of link counts in terms of finding the correct underlying model that lead to the observed data.
More recently, I have been venturing into the field of applied multivariate statistics. My focus lies in the visualisation of relationships between variables in high-dimensional data. I have given conference talks about the impact of data preparation on the possible results of dimension reduction methods and, most recently, an application of my work on avian community data gathered at bird feeders in New Zealand contributed to a journal publication.
Field of research codes
Applied Statistics (010401): Biostatistics (010402): Mathematical Sciences (010000): Statistics (010400): Stochastic Analysis and Modelling (010406)
Statistical modelling and inference for traffic networks
Visualisations and measures of association for multivariate data.
Project Title: Lattice polytope samplers: theory, methods and applications
Date Range: 2018 - 2021
Funding Bodies: Royal Society of New Zealand; University of Otago